Side note, but I hate that we're moving to a world where coding costs a subscription. I fell in love with coding because I could take my dad's old Thinkpad, install Linux for free - fire up Emacs and start hacking without an internet connection.
It doesn't though. You can still code the old fashioned way, and you are even likely to become a better programmer for it.
Personally, I tried copilot when I got it for free as a student and it didnt make a difference. The reason I know is that I was coding on two devices, one which had copilot installed the other didnt, and I didnt care enough to install it on the latter through an entire semester.
Its just slightly better autocomplete, by a questionable standard of "better".
> There’s literally nothing an llm can write or tell you that you can’t write yourself or find in a manual somewhere.
That's like saying, there's literally nothing a service business can do for you that you can't do yourself. It's only true in a theoretical sense, if neither time nor resources are a constraint.
In such hypothetical universe, you don't need a dentist - you only need to spend 5+ years in medical school + whatever extra it takes to become proficient with tools dentists use + whatever money it takes to buy that equipment. You also don't need accountants, lawyers, hairdressers, or construction companies. You can always learn this stuff and do it yourself better!
Truth is, time and attention is finite. Meanwhile, SOTA LLMs are cheap as dirt, they can do pretty much anything that involves text, and do it at the level of a mediocre specialist - i.e. they're literally better than you at anything except the few things you happen to be experienced in. Not perfect, by no means error-free - just better than you. I feel this still hasn't sunk in for most people.
> No, it’s not. You’re making my statement abstract for the sake of arguing.
> I’m not a cook, doctor, or a lawyer. I can’t prepare meals for a party of more than 2.
They are demonstrating how over-broad your own statement was with an *equivalent* statement to show how it only passes on an unhelpful technicality.
Immediately after your quotation is this:
> you only need to spend 5+ years in medical school + whatever extra it takes to become proficient
LLMs pass the bar exam and the medical exam. These are things which I assume I would be able to do myself if only I were willing to dedicate 5 years of my life to each.
> I can write exactly all code an llm could write.
I can often see many errors in the code that ChatGPT produces. Within my domain, it's just a speed-up, a first draft I have to fix. Outside my domain, it knows what I *can't* Google because I've never heard the keyword that would allow me to.
On legal questions, ChatGPT (despite passing the bar exam) seems to make up cases. I belive this because I can google the cases and fail to find them. Is this because they don't exist, or because they're not indexed on Google? I don't have the legal background necessary to know — and it would take me years to get the knowledge necessary to differentiate "it's worse than first glance" from "it's better than second glance".
Even if LLMs were "a memorization machine" (they're bad at that), the statement is obviously false because actutal literal memorisation machines (books, video recorders, Google) cannot pass these exams.
LLMs only started to because they could follow the questions.
But even that aside, it doesn't matter why LLMs can do what they can do or what else can also do that, what matters is that it would take most humans several years to get to the level of current LLMs in a subject that human isn't already familiar with.
I don't know how you compare, but ChatGPT's English grammar and vocabulary are significantly better than mine. And when I prompt it appropriately, it also seems to be a better creative writer than I am, at least for short pieces.
> Chatgpt (and other LLMs) are awful at creative prose.
As are most humans.
Don't get me wrong, what I've seen from even the better LLMs have a certain voice and tropes and sacherine worldview that isn't dark enough where it needs to be for the story to work; but on the other hand, what I see on some fiction writing subreddits… the AI is often a genuine improvement over amateur writers, even in cases where the AI contradicts itself about plot elements.
Which is frustrating, because I have the feeling the novel I've been trying to finish writing for the last decade may be usurped by AI before I get my final draft.
> what I see on some fiction writing subreddits… the AI is often a genuine improvement over amateur writers
What point are you trying to make here? That amateur writers are amateurs?
That AI is only "often" an improvement over an amateur?
> Which is frustrating, because I have the feeling the novel I've been trying to finish writing for the last decade may be usurped by AI before I get my final draft.
This statement shows such a warped attitude towards art and the creative process. What do you mean "usurped?" Do you actually believe that LLMs will overtake humans when it comes to creative works?
If so, you don't really understand what is compelling about the written word or what makes for good writing and reading and it's no wonder you feel as though your own writing is so substandard.
I highly doubt your writing is that bad. Especially if you've been working on it for a decade.
Just so we're clear, you're not sure what I'm saying and nevertheless think that I'm a better writer than you believe that I believe myself to be and you are confused by an imaginary "only" that wasn't in what I wrote and that you are skeptical of the idea that an LLM might overtake humans in creative works generally and writing in particular?
I'm not sure if the following statement will help your confusion, but most who judge the quality of a story do so without being able to write that story. Critiquing and writing are different skills.
It's giving a reply that "sounds right to a human" in context. Which, on small scale, is exactly what they, me and you are also doing when writing (or speaking), except for the infrequent cases when we force ourselves to reason through stuff very. slowly.
(This is why I believe LLM performance is best judged against human inner voice/system 1 reasoning, not the entirety of human thinking. When thinking with system 1, people don't really have an idea what they're doing either - they're just doing stuff that feels right.)
Also note that "sounds right to a human" is literally the loss function on which LLMs are trained, so between heaps of training inputs and subsequent extensive RLHF, the process is by its very construction aiming optimizing for above-human-average performance across the board.
Even documented libraries can be a struggle, especially if they are not particularly popular. I'm doing a project with WiFi/LoRa/MQTT on an ESP32. The WiFi code was fairly decent, but the MQTT and especially LoRa library code was nearly useless.
Sonnet 3.5 fails to generate basic JetpackCompose libraries properties properly. Maybe if somebody tried really hard to scrape all the documentation and force feed it, then it could work. But i don't if there are examples of this.
Like general LLM, but with complete Android/Kotlin pushed into it to fix the synapses.
Of course, why wouldn't it? It's a generative model, not a lookup table. Show it the library headers, and it'll give you decent results.
Obviously, if the library or code using it weren't part of the training data, and you don't supply either in the context of your request, then it won't generate valid code for it. But that's not LLM's fault.
Even if you absolutely have to use an LLM for some reason, there are already perfectly good LLMs for code generation that you can comfortably run on commodity hardware.
For now. It’s not clear what the monetisation strategy is, but probably it will be paid in future (alongside whatever other strategies they may have, like selling data, etc)
I think you are approaching this with the wrong mindset. I see it as I'm paying somebody to type and document for me. If you treat LLMs like a power tool, it is very easy to do a cost benefit analysis.
That is what happens when developers want to be paid for their work, but refuse to pay for the tools they use, regardless of little they may cost.
So we're going back to the last century, but given we are in a different computing context, only the stuff that can be gated via digital stores, or Web Services, gets to have a way to force people to pay.
You are not alone. The only future for these sorts of AI coding helpers is for them to use 100% free software AIs. On the bright side good progress is being made in that area and the main sticking point seems to be the expensive hardware to run them on (and integration). Costs on that hardware will hopefully drop over time so they won't still be mostly limited to 1st world (like the subscriptions).
But I am glad we now have more paid options available. Tooling is important and people that do good work should be able to charge for high quality tools.
I would be much happier in a world full of tools licensed like Sublime Text, where I can purchase a license and just run it without the need to constantly phone home though.
Use an aggregator like nano-gpt.com. You get access to all the top models (including o1 pro which usually requires a $200/month subscription) on a pay per use rate. Short on cash ? Use deepseek models for .1cents.
"install Linux for free - fire up Emacs and start hacking without an internet connection." that still works you know. Nobody is forcing you down this subscription path (except Microsoft)
Just get a graphics card and run a prompt-compatible llm yourself. Recent models like phi-4 show decent results (relative to your general amazement baseline) even on medium quantization. I’m running q4_k_m (8gb) with custom “just print and stfu” characters and rarely reach Claude anymore.
> I fell in love with coding because I could take my dad's old Thinkpad, install Linux for free - fire up Emacs and start hacking without an internet connection.
I'm not understanding what it is about a private company launching a product that changes that?
>but I hate that we're moving to a world where coding costs a subscription
I mean you don't need to if you don't want to. I am gainfully employed as a software developer and what I do everyday is literally just fire up Emacs on my Linux machine and write code. To this day I haven't figured out what llms are supposed to do that a bunch of yasnippets don't.
Just like five years ago most of my day is reading and debugging code, I'm not limited by how fast I can type.
Understand probably not, but they can read bizarrely fast and write a summary of what things do very fast ; maybe not 100% accurately but close enough to be of value when trying to understand a large bag of code fast.
Was going to say it's mostly been great for me. At least until recently trying to try out deepseek r1 it's not been great. I can't tell if it's the router or the provider..
If you pay out of pocket it means it's not an approved tool by your company, which means you can be fired and sued for leaking their intellectual property.
Also 20$ per month is way less than what it costs them to run it. Eventually they will need to charge way more to cover their costs, and the people who can't code without an AI assistant will need to pony up :)
It likely doesn’t matter if it is for training in this case. I can’t send our codebase to a friend even if they would never look at it or use it.
The issue would be based on the terms of employment and the software license. There’s likely a provision that just says “don’t share” regardless of what the other party will use it for.
To OP’s point, if your company is paying for the sub, then sending the codebase data would be an approved use of the codebase as part of your job.
I do a lot more work with cline and aider than without. For now that translates to a lot more money. In the end it's probably just going to be part of your normal job; you will just deliver a lot more in your 40 hour week than you did before. At this stage in the AI hype, I would insist on more or code side projects or do multiple projects. I write a lot more code and still have more time than before sonnet came out.
If it improves your performance by 1%, then for a salary of $2000/mo, that $20/mo is breakeven.
If it benefits you more than 1%, then you're in profit.
Of course, if you're in a job that doesn't actually care about performance, and performing won't lead to better salary at some point, then it may not matter.
Only if it increases your compensation by that 1%?
I get making economic/stats based analysis like this, but is your boss going to notice that 1% to give you a raise they otherwise wouldn’t? Probably not…
Your company culture can be performance-minded and this still be true.
What a weird way of thinking. If you act so pedantic anyway, you can stop coding simply 1 percent earlier each day, since your boss would not notice anyway. Personally I live in Belgium and I would be glad to pay 20 euros per month if it increases my productivity, as noticed by my boss or not.
The real problem however is, I cannot simply share my employer's repos to be absorbed by any LLM out there. So I use only the tools that my employer provides and approves of. Currently that is Microsoft Copilot chat/RAG via my work account. It takes some copy/paste and adaptations of problems/solutions but it is much more efficient than using SO. It is also a great teacher that never gets tired of my plenty why/how questions.
In my view, the future is that LLMs can train on entire private code repos until it understands its ins and outs. Currently it would need to fit in the context window, hence you need to babyspoon it, as I understand things.
I was only embodying the mindset in the comment I was replying to. Thinking in terms like “My time is worth $X, so if I do Y thing that takes Z minutes, that’s D dollars saved.” It gets really hairy and leads people down false paths where they think they’re being super “rational” all the time.
In this case it tricks you because it assumes that the LLMs increase productivity and launders that into the calculation. For me, and many people, LLM usage decreases my productivity.
I see many comments about not using this because China/ByteDance stealing data, etc. I get it for for enterprise, you don't want your code leaking to the outside world, even if in my experience, the code is mostly garbage and not worth that much.
But if it's for some personal project that you're putting on Github, does it matter? If my code is going on Github anyway, it's going to get slurped up regardless. I don't particularly care if it gets processed by Cursor or ByteDance before it gets scraped.
> But if it's for some personal project that you're putting on Github, does it matter?
I don't think the fear is that they'll steal code you'll end putting publicly on GitHub, but everything else. I guess there is some fear that it won't just analyze and process what you currently have open, but might scrape your computer for more data and so on.
I personally don't believe ByteDance would be stupid enough to even attempt exfiltrating files from developers machines, which typically are better protected than the average user computer, just trying to see the perspective of others with a more charitable reading :)
Desktop OS's typically don't protect one application's data from another, or the system's data from an application. At least not out of the box without configuring some kind of sandboxing solution.
The user-based permissions model is an outdated dinosaur from a time when we could trust the applications we run on our systems to act on our behalf. Applications now act on the developer's behalf, often against the user. An application "running as me" should not have access to every resource (file or peripheral) on the device that I have access to. That's a huge blast radius.
Operating Systems really need to start treating developers as adversarial from a security/permissions point of view.
Macs have started prompting for confirmation when software requests access to some directories. It’s better than nothing, but the OS doesn’t really have any kind of tutorial explaining how the permissions work and which directories are covered.
Trust em or not … had that happen to someone else using Cursor a few weeks back. I ran a prompt for refactoring a function and all of a sudden it started dumping someone else’s directory structure and file contents into my text file. Like hundreds of files. Not exactly sure what happened but I immediately uninstalled Cursor.
I'm glad people care about their personal data, I just wished they cared as much when talking about ChatGPT, because in reality if you're not a Chinese dissident living abroad, then OpenAI is much more dangerous to you than the CCP.
OpenAI does not have a history of disappearing people and when they return, their attitude has completely changed (see Jack Ma). OpenAI does not have a history of aggression against other nations shipping vessels. OpenAI does not have a history of corporate espionage and IP theft. OpenAI does not have a history of setting up police stations in other nations. OpenAI does not have a history of operating an illegal bio-lab in foreign nations. That's pretty defamatory to compare OpenAI to the CCP. OpenAI does not have a history of spreading a virus that killed millions and pointing fingers at other nations.
If we are talking countries here, then all that you've mentioned is done by both sides. If we are talking companies then as far as I know only one of them is developing killer robots (guess which one).
About the virus: you really need to be less in the company of people thinking up conspiracy theories. Remember that a pandemic has always been predicted in a 'not if but when' sense.
For the rest of the provided history feats, I can't really tell but most of it appeared indeed in established newschannels.
> […] in reality if you're not a Chinese dissident living abroad, then OpenAI is much more dangerous to you than the CCP.
Correct.
I laugh a little every time I hear some folks immediately push the “Big Bad China is just trying to steal our data” narrative. My immediate reaction is, “Grab some tea and let’s sit down real quick, so I can tell you what some of our companies right here in America and our government do with our data.”
Agreed. I get that both might have the same level of access to my data but if I was staying in USA, I would be much more okay with my data getting in the hands of an organization which is within the jurisdiction of the democratic country I live in. Also, given the global political scenario, China is considered a competitor/enemy due to opposite ideologies (not my personal opinion) so the choice is obvious. However, whether USA is/will stay a democracy for long is a longer debate.
Who cares what political stunts the US is currently going through? Except for the ~4% of the world who happens to live there, the rest of us tend to choose tools based on either what others around us use, or what we've found to work best for us.
If that happens to be an editor from China instead of US, I don't know what difference that would make? Both governments in those countries are crazy about spying on both their own citizens and everyone else, and have their corporations under surveillance.
As long as the editor doesn't send opened files/files from my drive to some remote backend, I couldn't care less about the nationality of the developers.
> As long as the editor doesn't send opened files/files from my drive to some remote backend, I couldn't care less about the nationality of the developers.
Is this sarcasm? Isn’t that exactly what the editor has to do to do its job? How else would the AI stuff work? It’s not running on-device, right?
I mean, I have some stuff from China, some of the stuff from there is cheap and high quality, so why wouldn't I? I don't know of many consumer products from Russia, North Korea or Iran. I'm sure if there was something from those countries that I really wanted, I'd buy it.
Edit: Actually I do have a product that is kind of Russian, Flipper Zero. I'm don't remember if I bought it before they moved the company to the US though, or while it was based in Russia.
Its not shut down, your president made it available again without a single change. And it was only a geoblock to begin with. By that logic it can't be that bad?
Not quite. At least in iOs, you can't redownload it, so if you uninstall it or change phones, no more TikTok. Also, Apple will be blocking any updates to the app for US folk.
TikTok was shut down because of clueless bureaucrats and political games, not because of any specific threat. If there was a real risk posed by China why are they idly allowing millions of people to migrate to Xiaohongshu, which is even more strongly controlled by China?
It was always BS though. AOC did an interview where she explains (what should be obvious to anyone), that US Gov had no proof whatsoever beyond trying to point at that Chinese law (much like Section 702 in the US), that makes TikTok seem risky.
Except that TikTok isn't legally a Chinese company, and isn't really under their jurisdiction.
Chinese companies like to be under the control of the Chinese government no more than big American companies like to be overly exposed to the US... Thus Apple being based in Ireland and keeping most of their profit outside of the US.
Absolutely and verifiably false, again! There are numerous data points that prove that the CCP exploits the TikTok algorithm to influence western views towards a pro-eastern bend. See my comment history for references. The close-doored national intelligence briefing that the senate got about the way the CCP manipulates the algorithm and on how TikTok is a national security threat resulted in a 100-0 vote in favor of banning TikTok. When do you get 100% of the senate to agree upon anything, ever?
TikTok was a threat, and will remain a threat until it’s divested from or shut down.
And the USA wants only fb and google to be able to tell them what propaganda to push. It's all a propaganda war, it's not about propaganda-vs-real news.
There's also the Section 702 angle... US can get all US social media to spy on non-US citizens without anything in the way of legal pushback.
US citizens have rights. Non-US citizens, do not.
I think the problem with TikTok under their audited, non-interference under their Project Texas is... That they might not easily hand over data, or suppress views that US Gov isn't a fan of.
Which will change once TikTok is owned by a US company.
They have all the US data on Oracle servers, with Oracle having the responsibility to monitor all access and make sure that no Chinese entity can get access.
All algorithms are monitored by Oracle such that only user actions can determine what they see.
So, unless you're claiming that "Oracle" is actively colluding with China, you're letting your worldview be defined by government propaganda.
The Pro-Palestine thing that TikTok refused to suppress at US Gov's command, is not "China promoting Eastern causes", it's "Americans" taking a position that the government wants suppressed.
Thus Biden being forced to cave and at least pretend that he cared about Palestinians towards the end of his Presidency... Which he stopped caring about once Democrats had lost and it no longer mattered what Democrat voters think for the time being.
We all know that's b.s. after the leaked audio recordings from TikTok meetings describing how US user data is still accessed from China.
> "Everything is seen in China," said a member of TikTok’s Trust and Safety department in a September 2021 meeting. In another September meeting, a director referred to one Beijing-based engineer as a "Master Admin" who "has access to everything."
The idea that Oracle can prevent Chinese employees from accessing data when the servers are running code written by ByteDance on bare metal and the US team literally reports to the Chinese one is, frankly, ridiculous.
TikTok ban plan may actually snowball, because now China thinking about their soft power and opening their internet to the world.
The 16th Shanghai Municipal People's Congress actually talked about this.
They have a lot of companies with very large consumer, and their market is saturated already. Time to expand to the rest of the world.
There's no way I am using a Chinese IDE with VSCode (w/ Coppilot) and Cursor existing after reading the privacy issues in the court documents with TikTok. Bytedance had some extremely overreaching expectations of data it wanted.
That being said, I am using (Chinese) Deepseek r1 because there isn't currently a free LLM on par with it. I am careful with what I share though, a little more so than with any others that are not locally fun.
I'm thinking about this the same way, if a bit more selfishly: as a non-US citizen, I might want to move to the US in the future.
Which is why I would like the US govt to have as little data as possible about me (who knows what the mainstream politics will look like, things that are very innocent today might be punishable by death in 10 years).
Chinese govt, on the other hand, can have my data freely, since I don't think I'll ever move to the Mainland China.
> (…) (who knows what the mainstream politics will look like, things that are very innocent today might be punishable by death in 10 years).
Maybe it’s just me, but I don’t think I’d ever consider moving to a country where I believed politics to be so unstable that an action could go from “very innocent” to “punishable by death” in the span of a decade.
Maybe it's just me but... personally I wouldn't move to a country in which more than half of the voters are pro a public jury-convicted criminal. I'm also reluctant to move to any country in which proper healthcare is reserved for the rich. I can't imagine moving to a country that is pro-life and pro-guns at the same time. A country that approves of, let alone encourages, getting rich by pump and dump of virtual freshly invented coins. Same sentiment i have about a country that approves of deportation of illegal immigrants, poor or rich, child or adult, criminal offender or not.
We’re talking about going from fully innocent and legal action to full on death penalty in just over two terms (depending on country). That seems quite a stretch for many countries where the death penalty was abolished.
It doesn’t seem plausible that in countries where even euthanasia isn’t allowed things could get so out of hand that the death penalty would not only be reimplemented but it would devolve so fast that you’d start executing people for things which aren’t even crimes right now.
Clearly the death penalty was not just an example, seeing as the reply from that same user made a good point about their own country becoming a war zone.
I recommend you stop assuming what other people meant and nitpicking for no discernible reason. The only one debating here is you, and you’re doing it with yourself.
Then you know nothing of how China operates its military or its intelligence gathering apparatus. Ask any of the parts of the world that disagree with China about their sovereignty…
...you realize the US is invited to be in those countries, right?
And why would you see enemy tanks on an island? What you do see are many many Chinese ships and planes, constantly violating Taiwanese air and sea boundaries.
Some of them exist against the will of the host country's government, for example in Cuba and Iraq, and Iraq's government is even democratically elected (albeit with issues)
That something else is even more toxic doesn't make poison not poison. It's not a choice between what poison you pick, but between accepting or rejecting poison.
A poison is not going to harm you unless it makes no contact with your body; what GP is saying is, regardless of which one is more potent, it's safer to pick the one you're least likely to ever find yourself near.
It's safer to not ingest any poison. You don't extinguish fire with gasoline because kerosene might be worse. You don't eat rat poison because fentanyl would be worse. It's a false dichotomy.
It's not a false dichotomy - the entire analogy to poison makes no sense, because US or China spying on people's activity on-line is currently a complete nothingburger. If anything, adtech is actually doing real harm to people this way, but that harm is still relatively small and indirect, which is why most people don't care.
> There's no way I am using a Chinese IDE with VSCode (w/ Coppilot) and Cursor existing after reading the privacy issues in the court documents with TikTok.
I have to say, on second reading, you are right because I basically overlooked that Copilot implies having no problem with being spied on by US corporations. That honestly escaped me.
But since you brought it up, for me mass surveillance and other abuses are simply a human rights issue more than anything. People can chose to not uphold theirs, or we can feel like we "have no other choice", but that doesn't change the human rights abuse, essentially. And we have no clue what the long term pay-off for everything being enmeshed into surveillance of every single data point, no matter how trivial in isolation it might be. And "I don't care about privacy, so let's not have privacy anymore, so nobody ever born will have privacy again" is not a decision anyone has the right to make, to put a point on it.
We all kinda ended up here, and I know it's not so easy, but that we don't have "fixed" it doesn't mean it's not broken, and it staying broken for a long time doesn't normalize it in my books. That people don't care is what raising awareness is for, after all.
And when people like Larry Ellison talk about everybody being on good behaviour because the AI will be always watching (to help everybody and make a great society, of course, because nothing could go wrong), basically, I think adtech isn't the only danger, and the effect of that would also not be "small and indirect". But it also underlines your point about what's the more direct threat to a US citizen.
And, logically - if we're worried about spying - it actually makes the most sense for average Chinese people to use US software and vice versa. The biggest threat to a person is always the police & military of their own country.
For example, China killed a lot of people. They were all Chinese.
I think you're unaware of just how aggressive China is. Look at their neighbours and their policies - they don't have a single friend because everyone is afraid of their bullying behaviour backed up by their military.
Look at conquered regions like Tibet and Xinjiang.
Look at how they set up police forces in foreign countries to keep an eye on Chinese citizens living abroad. Even having kidnapped and illegally held Chinese citizens in England because they posted anti-CCP messages on WeChat.
Look at countries like the Phillipines (not even a direct neighbour) who are trying to hold on to small fishing islands just off their coast because the CCP claimed those islands in the 1970s.
Remember a few years ago when they ran a week long military exercise around Taiwan... because a US representative spoke to the Taiwanese president. Sure I agree with you that the US also overstepped a line here, but for your response to be shelling the waters around the island is excessive to say the least.
Look at the aggressive nationalist and imperialist news they feed their own population, and the propaganda spread to make the Japanese seem like demons. Did you know there are several theme parks in China where children are encouraged to Bayonette a mannequin of Japanese Imperialist soldiers?
Call the US as bad as you like, but I've never seen a theme park where children are actively taught how to kill and demonise the Taliban or Nazis.
You could say all of that about the American empire.
We renamed it to the Gulf of America, talk about annexing Canada, own Hawaii, fund the colony of Israel, invented Manifest Destony, destabilize democracy via the CIA, firebombed Vietnam, etc etc etc
And don't talk to me abt Chinese spying when the NSA and Five Eyes hoovers up and processes the entire internet.
Re: propaganda, read your own comment. The American propaganda machine got you to demonize China pretty well
And to top it all off, despite having a much smaller population AMERICA HAS MORE PRISONERS THAN CHINA. Its insane!
In general I do agree (except for the recent Canada stuff - I believe it's more of a bluff than anything else)
But I want to respond to this:
> And to top it all off, despite having a much smaller population AMERICA HAS MORE PRISONERS THAN CHINA. Its insane
1. The US publically acknowledges who is in prison, China doesn't. You can be disappeared in China without anyone knowing.
2. The Chinese operate several Black prisons both within and without their borders. These are Prisons without actual sentences, laws or rights. The public doesn't even know of their existence. This is where you land up as a political prisoner in China.
3. The Chinese have placed thousands of Uigur muslims in "re-education camps" (not prisons) - many of whom are only guilty by association (i.e. there's no direct crime the CCP arrested them for, other than being a blood relative of someone who did commit one of these "crimes")
4. Chinese police are already spread thin enough dealing with all the political prisoners meaning there are many dangerous criminals who are freely committing crimes in China. The CCP keeps this information under tight wraps so as not to cause a panic. As a result the Chinese are a lot less cautious than they should be so (unofficial) crime levels are much higher than they should be. It's not uncommon to hear of (or see) Children who were kidnapped and had their limbs chopped off so they could beg more effectively - there has never been a widespread crackdown on this behaviour and many perpetrators are still walking the streets forcing children to beg for them.
5. Not to even talk about the prison organ harvesting claims made by several groups who have been persecuted by the CCP. From Falun Gong practitioners to Uigur muslims.
Also don't be lazy and just call this propaganda - I could easily do the same with other messages in this thread ("CCP propaganda") meaning it's not a strong argument to anyone on the fence / on the other side. Yes, it is very anti-China, but that's because I'm trying to bring it in contrast to the very anti-US message. I don't think either country is perfect and in fact I prefer to not live in (or near) either.
1 & 2: Wikipedia keeps an active list. Currently there are 15 held in Guantanamo bay. Could you provide me with a similar list of prisons + prisoners from China? (no because they intentionally keep it secret). I think the fact that we can name and put pressure on the US about this ~prison~ torture-camp makes enough of the argument for me. https://en.wikipedia.org/wiki/List_of_current_detainees_at_G...
4. No, and trying to use official sources unfortunately gives you an overly positive impression on China. This is one of the reasons it's so dangerous for an average citizen to live under a system which hides information - you are left without knowledge of how much danger you're in. If you think you're safe you're more likely to be taken advantage of.
There was an interesting case recently where CCP propaganda showed that other countries feared the Chinese passport (trying to boost nationalism and pride in China). This has led to several Chinese citizens getting robbed abroad, they then hold their passport in the attackers face thinking it will scare them off. Only for the passport to also get stolen.
6. Absolutely, the direct danger from the US isn't from the US itself but the governments it backs up (indirect danger?). Whether it's because the government is incompetent or evil, history shows that whatever the US props up is likely to collapse as soon as they leave but the country is in a much worse position than if the US never involved itself.
The US is largely talk, vs. the substantial action China takes and has taken against territories it considers its own.
The US importantly has a robust and independent judiciary where a fair trial is infinitely more likely than in China. Even for foreigners, the US legal system is accessible, understandable, and weildable for protection. It’s far from perfect, but 100x what you’d get in China.
Was Iraq also talk when US invaded it in violation of international law? Not to mention overthrowing democratically elected governments in South America.
Which countries did China invade in the last 3 decades?
I would like to highlight the small bit of positive in the American system. I don't support the removal of abortion law, nor do I like mixing religion with politics. But, due to the system that exists in the USA, even though the government removed the right to abortion, the individual states can still support it and form their own laws around it. However, a similar situation in other countries would result in complete removal without any option.
They got u there tho,
I'm with you that there's a ton of China hate deliberately stoked by US Govt and media, mostly as a distraction from out problems at home.
But there's nothing to defend about Tiananmen Square
No, cuz its flagged now, but the vibe I got was that the tank driver wasn't willing to run over the guy, and presumably thus wasn't so violent an incident as we make it out to be
I feel like the better move is to acknowledge that they have done and are doing atrocities but that its easier to stop the very similar ones we're doing than the ones being done halfway across the globe
They sent in the goons to beat up and arrest students protesting genocide or climate change too.
Hasn't gotten to tanks yet but truthfully it doesn't need to because we both know those protests aren't doing anything. If there was a credible student uprising it would happen here.
Taliban is a terrorist group, painting them in a bad light is factual.
Nazis, also factual.
So I don't think I fully understand the last sentence. Do you mean it's not appropriate for specifically _minors_ to have a theme park about how Nazis are bad guys? Because there's a lot of western content to this effect, including movies and video games, which are also accessible to children.
If anything, Chinese content guidelines usually prohibit graphical display of violence, so it's much more of a milquetoast thing than e.g. South Park.
This is fair and I think reflects my personal opinions on keeping children out of politics.
My point is more to do with seeing little kids being dressed up in CCP uniforms, handed a bayonette, and told to charge at the Japanese. I also agree we see this around the world "including movies and video games, which are also accessible to children" but I draw the line where it's being encouraged by the government to do these actions in person. Combine this with the anti-japanese rhetoric taught in primary education and it's a nasty combination*. [1]
Another thing to consider is that children are generally not given access to games/movies/whatever that have such mature themes by their parents.
I'd make this same argument if I saw US children being dressed up as US soldiers and asked to charge at Nazi "soldiers". Even though we can both (hopefully) agree that Nazi ideologies are/were disgusting and deserve to be bayonetted.
> If anything, Chinese content guidelines usually prohibit graphical display of violence, so it's much more of a milquetoast thing than e.g. South Park.
Not when the Japanese or another of China's enemies are involved. Then it's gloves off.
* Also I do want to give some slack here too, the Japanese have never acknowledged or apologised for their attrocities in WW2.
Fact check: China has strong ties with Central Asian states, Russia, NK, Myanmar, Laos, Cambodia, Pakistan. Regional tensions mainly involve US-allied nations.
Can you compare the military spending of one vs the other and report back please?
I never said china has never done anything bad. There is no such country on the planet.
China doing bad stuff doesn't make the bad stuff the other countries do disappear, and my point remains valid on which one is more of a threat to nationals of other countries.
I'm not making the claim that the US is innocent because China did all this stuff. I'm making the claim that I would feel much safer living on the Mexican-American, Canadian-American border or even in Panama, than I would near any of China's current conflicts. And that China is just as aggressive (if not more so) with their neighbours.
At least if the Americans win I don't have to be worried about becoming a political prisoner, forced into re-education camps, or having my organs harvested.
Also there are many more conflicts currently where China is involved than the US.
Living in these countries sounds to me like a nightmare (and I am from a 3rd world country). I’d prefer China because the ancient culture, history and thriving economy.
3rd world country or not - I hope you appreciate that you are incredibly fortunate for "ancient culture" and "history" to be 2 of your top 3 considerations for where to live and the quality of life that would give you.
Perhaps revalidate the "thriving economy" assertion before taking the plunge though (and be mindful of how hyper-localised that is).
I feel the need to point out that while many people might understand the point you're trying to make, the way this question is phrased doesn't do the best job of conveying it.
If I was presented with two options: waking up tomorrow as the child of a poor farmer in a third-world country, or waking up as one of Donald Trump's children, I would definitely choose the latter. However, that doesn't mean that I trust Trump more than I would trust the farmer. In other words, quality of life (or a preferred way of living) are not inherently tied to trust, morality, or anything like that.
Why did you choose to wake up as the child of a peasant in one country but the child of the president in another?
Quality of life is inherently tied to trust and morality - both in terms of the effect of fear (a lack of trust) or isolation (a lack of moral consensus, or equality and sense of shared belonging).
My question is not a false dilemma. As the link you gave points out, a false dilemma starts with a false premise which limits the subject to two absolutes when in-fact multiple exist. My question has no false premise, it's simply asking, if given this scenario, which would you choose and why. I am not saying that in reality you only have those options. The original comment setup the contrast of USA and China, so my question was a way to elucidate more about that contrast.
This is probably true, but at least in principle one of those foreign nations stands for free speech and democracy and the other stands for censorship and authoritarianism.
[Edit: hey Europeans commenting and downvoting below, note the words "in principle" in the above comment and evaluate which of the two countries do or do not purport to stand for these things despite whatever your hot take may be on the current moment.]
Huh? Does that negate the missing democracy, the political distrust and the political censorship within your country?
I am happy for you and your flags.
However I didn't claim that china is more "free" just that the US definitely isn't seen as that from the outside anymore. It's not always a direct competition
Given the last few days, As a EU citizen I'm not so sure such distinction is fully accurate anymore. And I'm pretty sure in 2 to 3 years from today the distinction will be even less noticeable.
I wouldn't trust a US data company to not capitulate to... "personal requests" by Musk or Trump any more than I trust it not happening in an hungarian, russian, turkish or chinese one with their respective leadership (official or otherwise).
That's actually funny.. considering that the main reason TikTok ban was proposed was the lack of censorship of sensitive topics like a certain genocide
I would rather let Chinese companies take my data than US/EU ones where I have bank accounts. Unless you are a super important, influential person, you shouldn't worry about China.
As far as I can tell this is operated out of Singapore, which is not some communist backwater, according to the law there and on top of US big tech infrastructure. The models this editor work with are by Anthropic and OpenAI both US companies.
Singapore or not, these TOS grant you nada in terms of privacy. You'd be a fool to use this: We may use Your Content to provide the Services to you and to other users, including without limitation troubleshooting, diagnostics, security and safety reviews, and customer support requests. You hereby grant to us, our affiliates and our third party partners (“SPRING Parties”) an unconditional, irrevocable, non-exclusive, royalty-free, sublicensable, transferable, perpetual and worldwide license, to reproduce, use, and modify Your Content in connection with the provision and improvement of the Services and its underlying technologies, as well as for the SPRING Parties' respective business operations, in each case, to the extent permitted by applicable laws.
One nuance to consider regarding Bytedance: TikTok’s CEO is also based in Singapore, but is a former Xiaomi executive who worked in Beijing for years. The firm’s employees in places like Singapore could be coerced by the government of China because of their friends, family, assets there.
Employees of Bytedance (and other Chinese companies) have to deal with draconian employee rules and agreements that tie them back to the rules of their Chinese mainland parent corporation and the Chinese government itself. For example look at the details that came out in this lawsuit against TikTok, where employees have to agree to uphold Chinese national interests, uphold socialism, etc.
I think you are contradicting yourself! Also, why add "China" to a product when you don't attach "America" to the other. If you are thinking politics, fine, else we are in for tech advancement!
Early TikTok was basically a standard Chinese startup in terms of how they collected data to optimize the app... It's not because they cared about what "you" are doing.
The Chinese tech ecosystem is just more competitive than the US one.
Most social media companies have been careless with data until they got public backlash though... TikTok's systems are probably more secure today than anyone else's... Because of the backlash.
Absolutely and verifiably false. WeChat became so important to the CCPs surveillance interests that it was essentially nationalized; and other apps that grow to its scale basically suffer the same fate. The only reason that hasn’t happened overtly to TikTok is because it’s an international product, so instead the nationalization happens discretely.
You don't hear about the other side because people like me usually get murdered in cold blood there is no difference but for you to believe that is a singularity like me being treated like a human :)
It is verifiably not controlled by the CCP... Refer Project Texas... And all US data already being on Oracle's cloud so that all access can be monitored (same for auditing of all their algorithms for any kind of programmed bias).
WeChat is an app that works inside China, so obviously it's subject to Chinese laws.
All countries have some restrictions on what is and is not okay to have on online platforms. China just adds in "social disruption" alongside content that could promote violence (which would be blocked on Facebook, etc. also)
You need to not just drink in propaganda... When something just "can't be done" (because it is being monitored), then it just doesn't matter even if you have ill intent.
Oh, so the Chinese governement is taking time from managing their huge country to stop that particular company from selling its recommendation algorithm.
China is taking every venture-funded AI startup and shitting on them. Almost all of them.
Tencent's Hunyuan 3D - Tripo, etc.
Tencent's Hunyuan Video - Sora, Runway, Pika, etc.
ByteDance's Trae - Cursor, etc.
DeepSeek R1 - GPT, etc.
Unitree - Figure, Tesla, Boston Dynamics, etc.
And a ton of this stuff is completely open source. All of the Tencent stuff is. This destroys the moat of so many companies that have spent hundreds of millions of dollars training and building their tech. It's just out there for free for anyone to build with.
And has anyone been checking the volume of Chinese AI research papers? Almost every impactful paper I've read in the last month has mostly Chinese names. And a lot of those papers come with code. Usually permissively licensed.
China is absolutely killing at the AI game. They've fast followed (or in many cases led) into a position of strength.
I wouldn't want to be RunwayML, Pika, Luma, or Tripo right now. Any "foundation model" company with a single use case is getting cloned and commoditized.
edit: Please don't downvote me because you don't like the message. I'm actually fine with you shooting the messenger, but this is absolutely worth talking about. It's pretty surreal to watch this all start to unfold over the last quarter or so. I want to read what others think about it.
Slightly relevant but they are also in the game of commercial mathematical optimization solvers and this year they reached parity with the sota American ones (ibm cplex, gurobi, Fico Xpress).
It is that bad that all the American companies withrew from the public benchmark library. You can only see there Chinese commercial solvers and open source ones.
Yes, I checked the recent solvers in the MIP space and it seems they're blowing out the existing American solvers. Huawei has invested a lot of money in this area.
But in my opinion this was completely expected, most research in this area is now done in China, American universities have in comparison stopped in time.
Any blogs or other writing about this topic you can recommend? I worked with gurobi in the past but haven't been keeping up with the trends and performance gains.
If these AI efforts are funded by the Chinese government, it makes perfect sense as a strategy. The more successful they are, the more risky it becomes to invest in US AI companies, leading to fewer investments (and less know-how in the long term), and China getting even more ahead.
(I am not in AI, so please correct me if this is inaccurate.)
Good news is the system in China frequently get in the way - people with connections wins the government or state owned businesses contract, making it tougher for many innovators to strive with a shrinking private-owned/consumer-demand environment
(Source: frustrated friends who lose the government contract to someone with inferior technology)
Trae doesn't look opensource, from their site looks like you download the binary (osx only for now) and install it. You can't configure it to use your own AI or another provider. You need to use their own API, all your prompts and code are go to bytedance.
Computer vision and deep learning conferences have been overwhelmed with Chinese authors since many years ago. No surprises. However the quality of the work skyrocketed recently
Interesting theory -- maybe trying to diluting/confusing the technology market with alternatives that phone home [to the LLM] with your content.
I just had to talk someone I know out of using RedNote as the TikTok shutdown loomed last week -- they had ZERO IDEA about its provenance. Unless you're paying attention, it's actually not clear what's what.
How could more competition be bad? Sure some startups' business plans no longer work and have to pivot but many more startups now have access to better products at lower cost.
1) They try similar like Zuck with LLama make sure GPT or Claude or Gemini is not the real winner in the West world like Android/iOS was for smartphone, Facebook/Whatsapp for social media/IM, Google Search for search, Windows for desktop OS. Majority people get used to to single product and they don't switch often.
2) Trying to kickstart community similar like LLama kickstarted big community around it that helped with tooling, testing, etc.
3) Slowing down development of ai companies in the west - giving them less data for training (after all midjourney, elevenlabs, llama kickstarted training on copyrighted content, it's pretty sure they using user data at least from e.g. free version of GPTo-mini) and bleeding their budget kind of war of attrition.
4) Familiarise the west audience with their ai models - after all still not many people using models like DeepkSeek since not many providers host them and majority of people don't have fast enough computer to run full models fast enough - Trae helps with that
Well, this is apparent if you follow the releases, but Silicon Valley doesn't want you to know that. The strategy of making this open source has a clear goal: to make it easier for other Chinese startups to use it.
Maybe they think it's in their national interest that a generation of Americans stops practicing their reading skills because of the tl;dr machine and stops building job skills that aren't reliant on the job-doing machine.
I'm okay with the software being made open to the public. I think ultimately it's good for society as a whole. The US is welcome to do the same (I.E. actually opening OpenAI) but I don't think the drawbacks of AI are mitigated by making it more profitable to the American establishment.
I feel like this would've been better as a VSCode extension. Copilot, Q, Gemini, all were able to take this approach. Also, VSCode isn't considered a full IDE and adding some AI features isn't enough to change that. It seems like they forked VS Code just for the ability to say they "created an IDE" in the same way other projects fork Chromium to "build a browser".
Nah the extension API is pretty limited. Copilot uses proprietary API’s not available to extensions.
If you really want an integrated experience, and not just a sidebar UI, you need to go the same route as Cursor and fork Code-OSS (the MIT-licensed part of VS Code, analogous to Chromium for Chrome)
What about Continue? It's an open source, bring your own api AI integration for vscode. It does everything that copilot does, including the editing-your-code-in-front-of-you diff style editor.
Reminds me of JetBrains AI, I think I'll stick to JetBrains solution for the time being. I'm not super crazy on AI coding solutions, but the JetBrains flavor hits a sweet spot for me, built-in and does what I usually need it to do.
It is definitely quirky, but I am able to use it mainly to scaffold things, and to ask it to refactor things. I only use it for personal projects so far where its no biggie if it doesn't work out, or to ask it about stack traces.
I've also blankly asked it if there's any issues with the current code, and found errors before I try to build / run a project.
Same here, I need to try it again but Copilot+Aider cover my needs right now. I would like LLM completion in my commit messages which Copilot doesn't have but JetBrains AI does IIRC. I really want JetBrains AI to get better, I think they are supposed to be adding Claude support this month? Or already did?
is it worth it to buy sourcegraph cody pro ($108/year) for claude 3.5 sonnet + "advanced contextual awareness and code intelligence", even though trae provides claude 3.5 sonnet for free now?
My personal experience has been that very basic stuff like converting an object to another shape or creating a simple validation function or things like that has been super effective with AI. I've gotten so lazy that I try and prompt it to exactly what I want instead of even editing a bit.
But the moment it gets a little less basic, its a different story. I absolutely hate the experience. So I love it and I hate it at the same time. Is it going to replace me? I don't feel like its there yet. And this has been the case for a while now.
There is just too much money and people invested in this that its hard to say anything negative about it. Nearly every VC has rebranded their websites around an AI-driven future. And, to be fair, it’s not a fad—it genuinely works well for many things. But for now, I’m still skeptical about how far it can really go.
According to the Discord, Claude 3.5 Sonnet and OpenAI GPT-4o. Also according to the Discord, "SPRING (SG) PTE. LTD. or one of its affiliates" is fully subsidizing the costs. This doesn't look like a bring your own keys type of deal.
If it's truly talking to OpenAI or Claude, then the traffic can be intercepted and redirected to your own AI router for other cloud AI providers or your local Ollama instance.
If you want the same thing but not as sketchy you can check out Cursor. You pay them, so it's a little more transparent what they're getting out of it.
Very unrelated, but: I run Linux on an older Thinkpad and I appreciate fast websites. This is a very well-designed website! It loads fast and it scrolls quite quickly.
Does anybody feel the same? I feel like nobody touched upon this, but it's always very nice to feel interfaces are responsive.
Interesting, I have the exact polar opposite perception.
I accessed this through a Qubes AppVM (no GPU, limited memory and CPU budget) and the presence of videos makes this a very slow scrolling experience for me.
In general, anything that involves JS/CSS animation/blur/effects makes sites pretty slow (up to unusable for me). The unlogged homepage for github.com for example, spins my cpu at 100%.
Not really, there' a world of difference between how every model I've tried¹ handles English vs my own native language². It's especially noticeable with smaller models, which produce coherent English, but completely break down in other languages: spewing out incorrect grammar, mixing words from other languages where they make zero sense, etc.
¹: which is dozens of them at this point, open and not.
Probably tough to fix small models with fine-tuning... You'd likely need to train them with enough data from languages (and maybe translations between them) that they need to support, so that they can connect ideas in different languages properly.
Separately, the fact that they can translate between languages where they have never seen any translation data in their inputs shows that they have internal models of the world and language, that go beyond what one might expect from a statistical parrot.
They do have world understanding - perhaps limited some by the fact that their input data may not cover a lot of everyday things.
Seems to have been heavily downvoted also, it's flown off the front page. Times have changes for HN. Also double standard when it comes to the like of Deepseek r1 earlier this week :shrug:
It’s free right now. For my use case worked and I like the interface. Will give it a try and see if AI features are better than ZED ones. Made a video here: https://youtu.be/5QOsVgumDiE
Also, anybody knows a decent IDE where AI is a first class citizen? So far my experience with plugins is, that they have too little context to be actually useful.
I know it's a marketing site so the speed is for the sake of the demo but the various animations showing a cursor scrolling past reams of text quicker than any human can scan, and just blindly hitting Accept makes for weird optics.
I wrote a Tetris game with it, but there’s a bug where the blocks suddenly disappear while falling. I haven’t been able to fix the bug, and I’m not sure if it’s an issue with the IDE or Sonnet. The experience isn’t very good.
Where does it say it is done by ByteDance? I would definitely use it it is really done by ByteDance... Couldn't find any information of ByteDance on the page though.
It's surprisingly very good. The AI responses are very good, and it's free to use. They don't charge anything for claude-3.5-sonnet. I also love its UI even though they totally copied JetBrains Fleet's UI.
> They don't charge anything for claude-3.5-sonnet
Makes you think about how they're extracting value from you to make up for this. I rack up non-significant amounts when using Claude 3.5 Sonnet for coding.
Considering the data they siphon from their other "free" apps, I don't want to think about what this does on developer machines with code bases and production access.
I took it for a spin and it looks great - a great way of playing with a LLM. Sure some of the elements are in IntelliJ, Copilot, or maybe Cursor (which I haven't tried) but it is all very well put together.
Recommended.
(But of course it would be nice to be able to run it against a local model which doesn't seem possible at the moment.)
You can't compare Zed with Vscode and it's cousins. I've used Zed myself and it is still very lacking, limited extensions.
Zed only has syntax highlighting and an ai assistant, but that's it. Even though Vscode (and the rest) is still also considered a text editor and not a IDE, they have by far way more features.
Sure, you get super duper high speed, but at what cost?
I've tried Zed and Ghostty and I just don't really get the speed thing. Typing latency is extremely unnoticeable in VS Code/Cursor and other Electron apps for me.
For me it's very noticeable. Typing, invoking lsp actions, opening terminal panel, switching between files. Everything feels much much faster. Switched to zed several months ago and never looked back
A disappointment as this is yet another wrapper around VSCode. It doesn't change/re-invent the current paradigms of writing code.
Effort would have been much better spent on getting AI auto-complete "merged" with LSP. But who is listening? Pretty much everyone these days think they can plug an LLM and have their life issues sorted out.
I feel like I'm taking crazy pills, because every time I try to use AI for any of my projects - including the "better" Anthropic and OpenAI models, I get terribly written, buggy code that has nothing to do with the domain or question. Unless I'm feeding it actual softball, single function questions it adds almost no value to my dev cycle. Mind you, all of this terrible code also costs a ton in tokens.
It's sad because I have seen some truly remarkable progress in LLMs, but I feel like we aren't allowed to be honest anymore that LLMs aren't going to replace programmers or moderate our expectations.
It takes some practice to get good using them. I use it as a sophisticated autocomplete. For example, if I'm building an api client layer in a web frontend, I might have `fooApi.ts`, `mockFooApi.ts`, and `barApi.ts`. instead of writing `mockBarApi.ts`, something that might take 5 - 15 minutes, I can feed those 3 existing files into context, and a few seconds and pennies later I have a nearly perfect file ready for review.
For any semi novel development I want to write it myself to establish the patterns I like and get an understanding of what I'm building. It's the boring stuff I hand over to Claude.
That's just it, I try not to write the boring parts at all. That's what libraries and design pattern are for - avoiding writing boilerplate.
I certainly can see it being useful for writing additional tests, although I am very worried based on what I've seen that it will introduce more bugs than it's worth, which is why it's still a tool I use for hobby projects and not for real work.
It looks like if the stuff you read online, written by people who need to optimize for various metrics of which objectivity is not among the top ones, shapes the perception of reality
I've done some stuff for a compiler in Java, backend web services in Python/Go/TS, frontend w/ React/TS.
I was a huge skeptic of this stuff at first but started using it about a year ago. I could definitely live without it, but it also saves me a significant amount of time.
I've been working on a feature in Python. With Copilot edits I just needed to find files the implemented the pattern, add it to the chat context. and write something like "implement feature x following the same pattern". It never gets it right the first time, but you can just keep the conversation going and have it iterate.
Afterwards I can just write /tests and Copilot generates reasonable tests. If it missed cases I can ask it to write cover those tests. Often times I can also just write literally "cover edge cases" and it handles all reasonable scenarios.
You're already above average if you can see that AI-generated code is bad. Those who are below average will think it's amazing and improving their abilities.
I feel like it's so hot or cold. People have been raving about Loveable "one-shotting" the creation of an app given one prompt. I tried to have it recreate a basic landing page from a screenshot and it wasn't even close. It invented some sections, reproduced others terribly and completely ignored others. I went through numerous prompts until I ran out for the day.
With what I was left with at that point, it probably would have been a wash to fix that code or just start from scratch manually.
I don't blame Loveable, I think it's a wonderful product bordering on magic, like v0 and Bolt.new. I just think they are amazing products that have been over hyped to god-like status.
I see this here often, but really, can someone please make a video or blog post with a complete code session so we can see what exactly happens? I am curious.
Can you provide a specific example of where an LLM failed? If you show us your prompt, your "want/need", and the result, we can better judge your situation. My guess: Your domain is weird/rare, so LLMs are terrible because their training data is very limited.
I am not a big fan of LLMs so I try them once in a while, asking to "implement blocked clause decomposition in Haskell." They can recite a paper (and several other papers, with references) almost verbatim, but they do not posess enough comprehension of what is going on in these papers. As time passes by, with each new LLM, the level of comprehension drops. Last time LLM tried to persuade me to write code instead of providing me with code.
-- Example usage:
let clause1 = Set.fromList [1, 2] -- represents (x1 ∨ x2)
let clause2 = Set.fromList [-1, 3] -- represents (¬x1 ∨ x3)
let formula = Set.fromList [clause1, clause2]
-- Decompose the formula
let (nonBlocked, blocked) = decompose formula
> My guess: Your domain is weird/rare, so LLMs are terrible because their training data is very limited.
The training data is always limited, because if there was an existing software that already did what I'm doing, I'd be using it instead of writing it :)
So by definition, unless I'm learning how to program and doing exercises that thousands have done before me, I'm doing something for which there is no training data.
It's great that you work on truly novel, unseen problems. The rest of us peasants tends to recombine previously solved problems into to novel solutions to solve business needs, and that works rather well with LLMs.
> The rest of us peasants tends to recombine previously solved problems into to novel solutions to solve business needs, and that works rather well with LLMs.
You took the words right out of my mouth! I wrote a draft reply to OP (but discarded it) with similar thoughts -- roughly: I work on CRUD apps, and so do most other devs; LLMs are a terrific fit for this subject matter.
It’s a tool with limited functionality. I think over time you vaguely learn the shape of its limits and work with what it can do, and it becomes genuinely useful.
Doing basic stuff with a very limited scope that’s likely well documented all over the internet, great.
Boilerplate, tedious yet simple things, works awesome as enhanced autocomplete.
More complex stuff, give it a shot, maybe you’ll get lucky, otherwise if it starts fucking up I’ve had the most success just taking a step back and doing it myself, maybe chatting with it as a live docs substitute, or a realtime stack overflow/discord programming channel which are also sometimes of dubious quality but frequently useful.
The misery really lies in getting stuck in that cycle of it just messing up over and over as you try to get it to make this thing work that is clearly beyond its scope and it’s just turning everything into a greater and greater mess of hallucinated bullshit.
> It’s a tool with limited functionality. I think over time you vaguely learn the shape of its limits and work with what it can do, and it becomes genuinely useful.
Yeah, I think this is exactly right. I use it for adding new files to an existing codebase, but without giving it access to that codebase, by passing in the definitions I want it to work with. It gets stuff wrong a lot, and I need to keep anything I'm asking for _well_ within a scope that I can be eagle-eyed about. But if I want to do something pretty simple that would be slightly annoying to write by hand, and can be done in a single file, it makes a pleasant alternative to typing out the code by hand.
I don't think I would be able to do this as a junior developer, I think this is only working because I can tell when it's full of shit, and I lasted about 90m of being willing to let IDE-integration happen, because it's too easy to be lazy and not scrutinize every little change, and that way lies total madness. This makes me beyond skeptical of non-developers writing anything significant with it: they'd be much, much better off learning Bubble or similar.
I think it's part art of prompting, and part problem scope -- they're not gonna get you a super polished final product without a LOT of support (yet), but they excel at rapid I-don't-care-what-the-code-looks-like-just-get-me-an-interactive-prototype work, and at solving specific problems (e.g. the function should do THIS, write it and write me a test suite). However, there are some agentic IDEs like Windsurf that are working to have LLMs orchestrate their own work. Well supported with a rock solid test suite, I've seen some people do pretty incredible things with them. The more a codebase provides tools for LLMs to live of the land in terms of tests and comments, the more aggressive you can be with what you ask them to do.
I've built a couple pretty simple apps that have been basically no code beyond tweaks for me.
I do think a big knowledge gap is the acceleration of prompting by knowing how to talk about code. LLMs show a lot of difference in response between "make the top bit stay on the screen on small screens" and "make the <th> element sticky for viewports <600px".
I'm pretty new in this, but a couple months back, I tried some direct AI code execution in Unity. It worked surprisingly well. Video: https://www.youtube.com/watch?v=0xCR4fiyugA
So I definitely think the value proposition of AI coding is higher when you have programmatical control over the workflow.
On a related note, the recent Cline Plan+Act function has also been a game changer.
As with all other tools you should expect there to be some learning curve. When desktop top publishing became a thing, professionals had to adjust their ways of working but if you stuck with glue and knives you were out of a job.
LLMs are a tool. It’s also weird and counterintuitive so it takes some time to get really good with them.
> I feel like we aren't allowed to be honest anymore that LLMs aren't going to replace programmers or moderate our expectations.
I haven't heard anyone in tech say programmers will be replaced by this tech.
I have heard persons outside of tech say it, but I think they lack sufficient context.
I mean, if we get AGI, yes, we might be fully out of a job. But, so far I think many programmers agree this seems to be a ways off.
But, I think you underestimate how many programmers used to copy paste code from StackOverflow and articles, etc.
We even had/have a phrase for that: "copy pasta".
Those programmers are getting more applicable templated code than they used to get via copy paste.
When I'm coding in my preferred languages, I am faster without AI.
But, when I'm writing yaml or something else for an unfamiliar tool or platform, I do get a productivity boost, even if I have to debug the code / configuration.
Bro the West is getting cooked by China right now. We might be fucked.
Everything that has been coming out of China is better than anything our mega trillion dollar corporation are producing and they are doing with with sanctions on the best GPUs.
I don't work in big tech, why are they so bad with sigifically more money and more advanced technology?
Google had a decade and a half head start and they can't even produce a model that is as good and as cheap as Deepseek building with no compute compared to Google.
I have been using Trae and VS Code with Copilot is laughable compare too it despite using the same models. So we can't even prompt LLMs as good as they do and they don't even speak English as a first language LMFAO.
We might be fucked and this is just the beginning.
Maybe time to learn code should be time to learn Chinese.
This won’t be a popular opinion here on HN but I think it’s the beginning of the end for Western hegemony. Asia is powering full steam ahead while we sit idle. We have much lower social cohesion and it’s getting worse all the time. We’re becoming lower trust societies. We don’t incentivise building enough. Success is considered having enough capital to sit back and generate more capital while contributing nothing back to society.
At the forefront of our politicians’ minds are how many genders there are. In the UK just building a new high speed railway has practically been the Manhattan project.
Will we get back on track? I think the chances of that are rapidly diminishing.
Well the west is too busy trying to answer the question “what is a woman” and allowing mass migration into their countries. While Asia is like “there’s men and woman, and we make it hard for you to live in our country and never get citizenship while we out build you in every aspect”
The chat UX is definitely better than Cursor. It less janky and I like that it wait until generation is done before asking you if you want to apply the suggested the changes instead of loading them into your file right after like cursor does. Also you can apply at cursor which is nice for snippets.
The Builder tool is probably the best feature of Trae. It builds React UIs better than any tool even v0 which is built for building React UIs. It very good at recreating UI from image even better than using the model directly which I'm not too sure how...
Code completion is probably closer to copilot than cursor so it depends on how hardcore you like your AI code completion. So Cursor might be better especially for repetitive refactoring.
The overall design and feel of Trae is better the VS Code and Cursor, but I'm a big fan of JetBrains and it feel a lot like a JetBrains IDE.
Does Jetbrains have a language agnostic IDE? Their page makes it look like you need to pick one specific to a language, but I have repos with multiple languages
DeepSeek is absolutely the leader in terms of cost value for high performance AI models.
US firms might have the lead in terms of absolute best performance, true. But their prices are 10-20x more for a product that is marginally better at most. When it comes to coding DeepSeek V3 is very close to Claude 3.5 Sonnet and exceeds GPT-4o.
I can't imagine getting an engineering job at an engineering company and asking them if I'd be allowed to create an opensource IDE that would never be a sold product. I can barely even get allowed time to write SRE tooling to improve our own engineers lives.
Some C-levels newphew into computers and need a school project or something?
Bytedance created all internal tools by themselves. Personally I'd definitely love to join such a company if I'm in that tooling team. Whether they are being used by outsiders is irrelevant because I already bagged the skills.
Eh, I make tooling for a living. It sucks to have nobody using your tools even if they were fun to make. You spend months/years putting something out there that you think will really benefit a company/team/foss and it goes nowhere. It's like painting a painting you love and it getting thrown in an attic.
We're truly building walls everywhere.
Personally, I tried copilot when I got it for free as a student and it didnt make a difference. The reason I know is that I was coding on two devices, one which had copilot installed the other didnt, and I didnt care enough to install it on the latter through an entire semester.
Its just slightly better autocomplete, by a questionable standard of "better".
There’s literally nothing an llm can write or tell you that you can’t write yourself or find in a manual somewhere.
That's like saying, there's literally nothing a service business can do for you that you can't do yourself. It's only true in a theoretical sense, if neither time nor resources are a constraint.
In such hypothetical universe, you don't need a dentist - you only need to spend 5+ years in medical school + whatever extra it takes to become proficient with tools dentists use + whatever money it takes to buy that equipment. You also don't need accountants, lawyers, hairdressers, or construction companies. You can always learn this stuff and do it yourself better!
Truth is, time and attention is finite. Meanwhile, SOTA LLMs are cheap as dirt, they can do pretty much anything that involves text, and do it at the level of a mediocre specialist - i.e. they're literally better than you at anything except the few things you happen to be experienced in. Not perfect, by no means error-free - just better than you. I feel this still hasn't sunk in for most people.
No, it’s not. You’re making my statement abstract for the sake of arguing.
I’m not a cook, doctor, or a lawyer. I can’t prepare meals for a party of more than 2.
I can’t perform surgery.
I can’t effectively defend myself in a court of law.
I (and I assume OP) have programming expertise.
I can write exactly all code an llm could write.
For simple scripts, demos and other easily Googleable tasks, LLMs will be faster, but it’s nothing out of reach for me.
These tools won’t force you to pay a subscription to code. You don’t need them if you already have experience.
> I’m not a cook, doctor, or a lawyer. I can’t prepare meals for a party of more than 2.
They are demonstrating how over-broad your own statement was with an *equivalent* statement to show how it only passes on an unhelpful technicality.
Immediately after your quotation is this:
> you only need to spend 5+ years in medical school + whatever extra it takes to become proficient
LLMs pass the bar exam and the medical exam. These are things which I assume I would be able to do myself if only I were willing to dedicate 5 years of my life to each.
> I can write exactly all code an llm could write.
I can often see many errors in the code that ChatGPT produces. Within my domain, it's just a speed-up, a first draft I have to fix. Outside my domain, it knows what I *can't* Google because I've never heard the keyword that would allow me to.
On legal questions, ChatGPT (despite passing the bar exam) seems to make up cases. I belive this because I can google the cases and fail to find them. Is this because they don't exist, or because they're not indexed on Google? I don't have the legal background necessary to know — and it would take me years to get the knowledge necessary to differentiate "it's worse than first glance" from "it's better than second glance".
LLMs only started to because they could follow the questions.
But even that aside, it doesn't matter why LLMs can do what they can do or what else can also do that, what matters is that it would take most humans several years to get to the level of current LLMs in a subject that human isn't already familiar with.
If your job is to write software, then you're the accountant or lawyer or doctor. Otherwise, what do we even bring to the table?
Like speaking english and coding?
Don’t be so hard on yourself.
Chatgpt (and other LLMs) are awful at creative prose.
As are most humans.
Don't get me wrong, what I've seen from even the better LLMs have a certain voice and tropes and sacherine worldview that isn't dark enough where it needs to be for the story to work; but on the other hand, what I see on some fiction writing subreddits… the AI is often a genuine improvement over amateur writers, even in cases where the AI contradicts itself about plot elements.
Which is frustrating, because I have the feeling the novel I've been trying to finish writing for the last decade may be usurped by AI before I get my final draft.
What point are you trying to make here? That amateur writers are amateurs? That AI is only "often" an improvement over an amateur?
> Which is frustrating, because I have the feeling the novel I've been trying to finish writing for the last decade may be usurped by AI before I get my final draft.
This statement shows such a warped attitude towards art and the creative process. What do you mean "usurped?" Do you actually believe that LLMs will overtake humans when it comes to creative works?
If so, you don't really understand what is compelling about the written word or what makes for good writing and reading and it's no wonder you feel as though your own writing is so substandard.
I highly doubt your writing is that bad. Especially if you've been working on it for a decade.
I'm not sure if the following statement will help your confusion, but most who judge the quality of a story do so without being able to write that story. Critiquing and writing are different skills.
(This is why I believe LLM performance is best judged against human inner voice/system 1 reasoning, not the entirety of human thinking. When thinking with system 1, people don't really have an idea what they're doing either - they're just doing stuff that feels right.)
Also note that "sounds right to a human" is literally the loss function on which LLMs are trained, so between heaps of training inputs and subsequent extensive RLHF, the process is by its very construction aiming optimizing for above-human-average performance across the board.
Also, local llms with an agentic tool can be a lot of fun to quickly prototype things. Quality can be hit or miss.
Hopefully the work trickles down to local models long-term.
Obviously, if the library or code using it weren't part of the training data, and you don't supply either in the context of your request, then it won't generate valid code for it. But that's not LLM's fault.
You can imagine the classic attention mechanism as a lookup table, actually.
Transformers are layers and layers and layers of lookup tables.
I use that with avante.vim for tedious refactors. All local.
I think you are approaching this with the wrong mindset. I see it as I'm paying somebody to type and document for me. If you treat LLMs like a power tool, it is very easy to do a cost benefit analysis.
So we're going back to the last century, but given we are in a different computing context, only the stuff that can be gated via digital stores, or Web Services, gets to have a way to force people to pay.
But I am glad we now have more paid options available. Tooling is important and people that do good work should be able to charge for high quality tools.
I would be much happier in a world full of tools licensed like Sublime Text, where I can purchase a license and just run it without the need to constantly phone home though.
Nothing stopping you to build the world you want really.
There's no moat, all the clever prompting tricks Cursor et al. are just that - there is no secret sauce besides the model at the other end.
Complexity isn't an issue either, have the model write the interface to itself.
I'm not understanding what it is about a private company launching a product that changes that?
You can do it without IDEs, nothing is stopping you. I don't think this is a new phenomenon though.
You are free to coding without spend a dime, these AI dev tool cost money because these LLM cost money to run
You can get the same experience with open source tools that you can run your own model on your pc
I mean you don't need to if you don't want to. I am gainfully employed as a software developer and what I do everyday is literally just fire up Emacs on my Linux machine and write code. To this day I haven't figured out what llms are supposed to do that a bunch of yasnippets don't.
Just like five years ago most of my day is reading and debugging code, I'm not limited by how fast I can type.
It is kind of terrifying that I probably would stop coding for the day if those subscriptions end. (I get far too much convenience out of them)
I have tried to rationalize it by the fact that I do pay for internet, and version control, and my peripherals etc
The problem is that coding was a passion, but turned out to be very lucrative profession so loads of people who can't do it want to do it.
This is why we have languages like Go, and AI tools: allow people who don't want to learn how to be developers, to get a job as developers.
Also 20$ per month is way less than what it costs them to run it. Eventually they will need to charge way more to cover their costs, and the people who can't code without an AI assistant will need to pony up :)
That only applies to regular ChatGPT use.
And developers actively using AI for coding can easily spend more than $20/mo for the API.
There are people spending $10-15/day in OpenAI API usage working through Cline.
The issue would be based on the terms of employment and the software license. There’s likely a provision that just says “don’t share” regardless of what the other party will use it for.
To OP’s point, if your company is paying for the sub, then sending the codebase data would be an approved use of the codebase as part of your job.
I hope your current and future employers never find this side of your personality :D
If you’re not running the model locally, you’re sending your code to them for analysis. Now ByteDance has it.
If it benefits you more than 1%, then you're in profit.
Of course, if you're in a job that doesn't actually care about performance, and performing won't lead to better salary at some point, then it may not matter.
I get making economic/stats based analysis like this, but is your boss going to notice that 1% to give you a raise they otherwise wouldn’t? Probably not…
Your company culture can be performance-minded and this still be true.
The real problem however is, I cannot simply share my employer's repos to be absorbed by any LLM out there. So I use only the tools that my employer provides and approves of. Currently that is Microsoft Copilot chat/RAG via my work account. It takes some copy/paste and adaptations of problems/solutions but it is much more efficient than using SO. It is also a great teacher that never gets tired of my plenty why/how questions.
In my view, the future is that LLMs can train on entire private code repos until it understands its ins and outs. Currently it would need to fit in the context window, hence you need to babyspoon it, as I understand things.
In this case it tricks you because it assumes that the LLMs increase productivity and launders that into the calculation. For me, and many people, LLM usage decreases my productivity.
https://github.com/kelseyhightower/nocode
https://github.com/getcursor/cursor
https://github.com/Trae-AI/Trae
Not even a commit: https://github.com/timdorr/-/commit/9e5a571abd3fc4f8714e8c40...
This is what a commit-less repo looks like on GH: https://github.com/verdverm/_
Unfortunately they don't show the description "starting repository for any language and project" (a play on the _ (any) token in CUE)
But if it's for some personal project that you're putting on Github, does it matter? If my code is going on Github anyway, it's going to get slurped up regardless. I don't particularly care if it gets processed by Cursor or ByteDance before it gets scraped.
I don't think the fear is that they'll steal code you'll end putting publicly on GitHub, but everything else. I guess there is some fear that it won't just analyze and process what you currently have open, but might scrape your computer for more data and so on.
I personally don't believe ByteDance would be stupid enough to even attempt exfiltrating files from developers machines, which typically are better protected than the average user computer, just trying to see the perspective of others with a more charitable reading :)
The user-based permissions model is an outdated dinosaur from a time when we could trust the applications we run on our systems to act on our behalf. Applications now act on the developer's behalf, often against the user. An application "running as me" should not have access to every resource (file or peripheral) on the device that I have access to. That's a huge blast radius.
Operating Systems really need to start treating developers as adversarial from a security/permissions point of view.
For the rest of the provided history feats, I can't really tell but most of it appeared indeed in established newschannels.
Correct.
I laugh a little every time I hear some folks immediately push the “Big Bad China is just trying to steal our data” narrative. My immediate reaction is, “Grab some tea and let’s sit down real quick, so I can tell you what some of our companies right here in America and our government do with our data.”
If that happens to be an editor from China instead of US, I don't know what difference that would make? Both governments in those countries are crazy about spying on both their own citizens and everyone else, and have their corporations under surveillance.
As long as the editor doesn't send opened files/files from my drive to some remote backend, I couldn't care less about the nationality of the developers.
I see your point but in this particular case, I think it needs to send the opened files to a backend for processing
Is this sarcasm? Isn’t that exactly what the editor has to do to do its job? How else would the AI stuff work? It’s not running on-device, right?
Edit: Actually I do have a product that is kind of Russian, Flipper Zero. I'm don't remember if I bought it before they moved the company to the US though, or while it was based in Russia.
(I am guessing that there will be a deal floated for Elon to buy TikTok in a few weeks)
Although, if TikTok's earnings can cover the payments on the loan, it could be possible.
Apparently, the executive order was illegal
It was banned because of a law passed by Congress. Representatives aren't bureaucrats.
Except that TikTok isn't legally a Chinese company, and isn't really under their jurisdiction.
Chinese companies like to be under the control of the Chinese government no more than big American companies like to be overly exposed to the US... Thus Apple being based in Ireland and keeping most of their profit outside of the US.
TikTok was a threat, and will remain a threat until it’s divested from or shut down.
US citizens have rights. Non-US citizens, do not.
I think the problem with TikTok under their audited, non-interference under their Project Texas is... That they might not easily hand over data, or suppress views that US Gov isn't a fan of.
Which will change once TikTok is owned by a US company.
It's funny seeing things like US Gov trying to control free speech, while being restricted by the Constitution.
They do still manage via back-channels.
All algorithms are monitored by Oracle such that only user actions can determine what they see.
So, unless you're claiming that "Oracle" is actively colluding with China, you're letting your worldview be defined by government propaganda.
The Pro-Palestine thing that TikTok refused to suppress at US Gov's command, is not "China promoting Eastern causes", it's "Americans" taking a position that the government wants suppressed.
Thus Biden being forced to cave and at least pretend that he cared about Palestinians towards the end of his Presidency... Which he stopped caring about once Democrats had lost and it no longer mattered what Democrat voters think for the time being.
> "Everything is seen in China," said a member of TikTok’s Trust and Safety department in a September 2021 meeting. In another September meeting, a director referred to one Beijing-based engineer as a "Master Admin" who "has access to everything."
The idea that Oracle can prevent Chinese employees from accessing data when the servers are running code written by ByteDance on bare metal and the US team literally reports to the Chinese one is, frankly, ridiculous.
https://www.buzzfeednews.com/article/emilybakerwhite/tiktok-...
They have a lot of companies with very large consumer, and their market is saturated already. Time to expand to the rest of the world.
That being said, I am using (Chinese) Deepseek r1 because there isn't currently a free LLM on par with it. I am careful with what I share though, a little more so than with any others that are not locally fun.
So as a non USA citizen, I'm way more afraid of USA than I am of China.
Which is why I would like the US govt to have as little data as possible about me (who knows what the mainstream politics will look like, things that are very innocent today might be punishable by death in 10 years).
Chinese govt, on the other hand, can have my data freely, since I don't think I'll ever move to the Mainland China.
> (…) (who knows what the mainstream politics will look like, things that are very innocent today might be punishable by death in 10 years).
Maybe it’s just me, but I don’t think I’d ever consider moving to a country where I believed politics to be so unstable that an action could go from “very innocent” to “punishable by death” in the span of a decade.
(I also understand that my outsider view on the US political system isn't well-informed. Still, better safe than sorry.)
https://en.wikipedia.org/wiki/Capital_punishment_by_country
It doesn’t seem plausible that in countries where even euthanasia isn’t allowed things could get so out of hand that the death penalty would not only be reimplemented but it would devolve so fast that you’d start executing people for things which aren’t even crimes right now.
https://news.ycombinator.com/item?id=42803165
I recommend you stop assuming what other people meant and nitpicking for no discernible reason. The only one debating here is you, and you’re doing it with yourself.
What I see is this: https://en.wikipedia.org/wiki/List_of_countries_with_oversea...
As I've already said, I never claimed china are the good guys. Just that they aren't the worse guys.
And why would you see enemy tanks on an island? What you do see are many many Chinese ships and planes, constantly violating Taiwanese air and sea boundaries.
I have to say, on second reading, you are right because I basically overlooked that Copilot implies having no problem with being spied on by US corporations. That honestly escaped me.
But since you brought it up, for me mass surveillance and other abuses are simply a human rights issue more than anything. People can chose to not uphold theirs, or we can feel like we "have no other choice", but that doesn't change the human rights abuse, essentially. And we have no clue what the long term pay-off for everything being enmeshed into surveillance of every single data point, no matter how trivial in isolation it might be. And "I don't care about privacy, so let's not have privacy anymore, so nobody ever born will have privacy again" is not a decision anyone has the right to make, to put a point on it.
We all kinda ended up here, and I know it's not so easy, but that we don't have "fixed" it doesn't mean it's not broken, and it staying broken for a long time doesn't normalize it in my books. That people don't care is what raising awareness is for, after all.
And when people like Larry Ellison talk about everybody being on good behaviour because the AI will be always watching (to help everybody and make a great society, of course, because nothing could go wrong), basically, I think adtech isn't the only danger, and the effect of that would also not be "small and indirect". But it also underlines your point about what's the more direct threat to a US citizen.
For example, China killed a lot of people. They were all Chinese.
Look at conquered regions like Tibet and Xinjiang.
Look at how they set up police forces in foreign countries to keep an eye on Chinese citizens living abroad. Even having kidnapped and illegally held Chinese citizens in England because they posted anti-CCP messages on WeChat.
Look at countries like the Phillipines (not even a direct neighbour) who are trying to hold on to small fishing islands just off their coast because the CCP claimed those islands in the 1970s.
Remember a few years ago when they ran a week long military exercise around Taiwan... because a US representative spoke to the Taiwanese president. Sure I agree with you that the US also overstepped a line here, but for your response to be shelling the waters around the island is excessive to say the least.
Look at the aggressive nationalist and imperialist news they feed their own population, and the propaganda spread to make the Japanese seem like demons. Did you know there are several theme parks in China where children are encouraged to Bayonette a mannequin of Japanese Imperialist soldiers?
Call the US as bad as you like, but I've never seen a theme park where children are actively taught how to kill and demonise the Taliban or Nazis.
We renamed it to the Gulf of America, talk about annexing Canada, own Hawaii, fund the colony of Israel, invented Manifest Destony, destabilize democracy via the CIA, firebombed Vietnam, etc etc etc
And don't talk to me abt Chinese spying when the NSA and Five Eyes hoovers up and processes the entire internet.
Re: propaganda, read your own comment. The American propaganda machine got you to demonize China pretty well
And to top it all off, despite having a much smaller population AMERICA HAS MORE PRISONERS THAN CHINA. Its insane!
But I want to respond to this:
> And to top it all off, despite having a much smaller population AMERICA HAS MORE PRISONERS THAN CHINA. Its insane
1. The US publically acknowledges who is in prison, China doesn't. You can be disappeared in China without anyone knowing.
2. The Chinese operate several Black prisons both within and without their borders. These are Prisons without actual sentences, laws or rights. The public doesn't even know of their existence. This is where you land up as a political prisoner in China.
3. The Chinese have placed thousands of Uigur muslims in "re-education camps" (not prisons) - many of whom are only guilty by association (i.e. there's no direct crime the CCP arrested them for, other than being a blood relative of someone who did commit one of these "crimes")
4. Chinese police are already spread thin enough dealing with all the political prisoners meaning there are many dangerous criminals who are freely committing crimes in China. The CCP keeps this information under tight wraps so as not to cause a panic. As a result the Chinese are a lot less cautious than they should be so (unofficial) crime levels are much higher than they should be. It's not uncommon to hear of (or see) Children who were kidnapped and had their limbs chopped off so they could beg more effectively - there has never been a widespread crackdown on this behaviour and many perpetrators are still walking the streets forcing children to beg for them.
5. Not to even talk about the prison organ harvesting claims made by several groups who have been persecuted by the CCP. From Falun Gong practitioners to Uigur muslims.
Also don't be lazy and just call this propaganda - I could easily do the same with other messages in this thread ("CCP propaganda") meaning it's not a strong argument to anyone on the fence / on the other side. Yes, it is very anti-China, but that's because I'm trying to bring it in contrast to the very anti-US message. I don't think either country is perfect and in fact I prefer to not live in (or near) either.
3. Yeah it's sad. You can find a list of police files which have been confirmed by 3rd parties. Based on 2018 data we have >5,000 in detention currently. https://www.icij.org/investigations/china-cables/xinjiang-po...
4. No, and trying to use official sources unfortunately gives you an overly positive impression on China. This is one of the reasons it's so dangerous for an average citizen to live under a system which hides information - you are left without knowledge of how much danger you're in. If you think you're safe you're more likely to be taken advantage of.
There was an interesting case recently where CCP propaganda showed that other countries feared the Chinese passport (trying to boost nationalism and pride in China). This has led to several Chinese citizens getting robbed abroad, they then hold their passport in the attackers face thinking it will scare them off. Only for the passport to also get stolen.
6. Absolutely, the direct danger from the US isn't from the US itself but the governments it backs up (indirect danger?). Whether it's because the government is incompetent or evil, history shows that whatever the US props up is likely to collapse as soon as they leave but the country is in a much worse position than if the US never involved itself.
The US importantly has a robust and independent judiciary where a fair trial is infinitely more likely than in China. Even for foreigners, the US legal system is accessible, understandable, and weildable for protection. It’s far from perfect, but 100x what you’d get in China.
Which countries did China invade in the last 3 decades?
Except if you have an abortion. Then laws don't matter.
They already started this work yesterday: https://www.politico.com/live-updates/2025/01/22/congress/se...
You went from “eh, China isn’t so bad” to “Tiananmen Square protestors had it coming” within one thread.
You people with no morals try really really hard to twist words uh? But at the end of the day you're just a sad racist.
I’ll save you time: you won’t find one.
Have you no capability of reasoning?
But there's nothing to defend about Tiananmen Square
I feel like the better move is to acknowledge that they have done and are doing atrocities but that its easier to stop the very similar ones we're doing than the ones being done halfway across the globe
So now your accusation is based on "bad vibe"… Pretty in character for someone not interested in facts :)
The proper move would be to show the very similar US response to Students Protests - https://en.m.wikipedia.org/wiki/Kent_State_shootings
They sent in the goons to beat up and arrest students protesting genocide or climate change too.
Hasn't gotten to tanks yet but truthfully it doesn't need to because we both know those protests aren't doing anything. If there was a credible student uprising it would happen here.
Nazis, also factual.
So I don't think I fully understand the last sentence. Do you mean it's not appropriate for specifically _minors_ to have a theme park about how Nazis are bad guys? Because there's a lot of western content to this effect, including movies and video games, which are also accessible to children.
If anything, Chinese content guidelines usually prohibit graphical display of violence, so it's much more of a milquetoast thing than e.g. South Park.
My point is more to do with seeing little kids being dressed up in CCP uniforms, handed a bayonette, and told to charge at the Japanese. I also agree we see this around the world "including movies and video games, which are also accessible to children" but I draw the line where it's being encouraged by the government to do these actions in person. Combine this with the anti-japanese rhetoric taught in primary education and it's a nasty combination*. [1]
Another thing to consider is that children are generally not given access to games/movies/whatever that have such mature themes by their parents.
I'd make this same argument if I saw US children being dressed up as US soldiers and asked to charge at Nazi "soldiers". Even though we can both (hopefully) agree that Nazi ideologies are/were disgusting and deserve to be bayonetted.
> If anything, Chinese content guidelines usually prohibit graphical display of violence, so it's much more of a milquetoast thing than e.g. South Park.
Not when the Japanese or another of China's enemies are involved. Then it's gloves off.
* Also I do want to give some slack here too, the Japanese have never acknowledged or apologised for their attrocities in WW2.
[1] https://asiatimes.com/2024/07/china-scrambling-to-unplug-ant...
"(A) learn that communism has led to the deaths of over 100,000,000 victims worldwide;
(B) understand the dangers of communism and similar political ideologies; and
(C) understand that 1,500,000,000 people still suffer under communism."
Including an oral history "Portraits in Patriotism"
If this isn't propaganda against the CCP I don't know what is. Call it truthful or not, kids are being told China is The Enemy.
They're dressed up in military uniforms as early as middle school if you're in JROTC.
And funny you bring up them suppressing hate speech when here in America we've recently decided to do the opposite with X and Facebook.
I never said china has never done anything bad. There is no such country on the planet.
China doing bad stuff doesn't make the bad stuff the other countries do disappear, and my point remains valid on which one is more of a threat to nationals of other countries.
At least if the Americans win I don't have to be worried about becoming a political prisoner, forced into re-education camps, or having my organs harvested.
Also there are many more conflicts currently where China is involved than the US.
I live in scandinavia and I can tell you people aren't feeling particularly safe about USA not going to invade.
(I don't think they'll do it for real… but then again I also didn't think Putin would go into Ukraine for real so, don't take my work as gospel)
Perhaps revalidate the "thriving economy" assertion before taking the plunge though (and be mindful of how hyper-localised that is).
If I was presented with two options: waking up tomorrow as the child of a poor farmer in a third-world country, or waking up as one of Donald Trump's children, I would definitely choose the latter. However, that doesn't mean that I trust Trump more than I would trust the farmer. In other words, quality of life (or a preferred way of living) are not inherently tied to trust, morality, or anything like that.
Quality of life is inherently tied to trust and morality - both in terms of the effect of fear (a lack of trust) or isolation (a lack of moral consensus, or equality and sense of shared belonging).
[Edit: hey Europeans commenting and downvoting below, note the words "in principle" in the above comment and evaluate which of the two countries do or do not purport to stand for these things despite whatever your hot take may be on the current moment.]
Edit:// "In principle" chinas gov stands for stability, economic development and national unity...
I am happy for you and your flags.
However I didn't claim that china is more "free" just that the US definitely isn't seen as that from the outside anymore. It's not always a direct competition
As did Stalin, Mussolini, Hirohito and the National Socialist Party. It's the "how" that matters.
Sacha Baron Cohen's The Dictator is a piece of art that demonstrates this absurdity in the best possible way.
Only of of them is hypocritical, that's it.
I wouldn't trust a US data company to not capitulate to... "personal requests" by Musk or Trump any more than I trust it not happening in an hungarian, russian, turkish or chinese one with their respective leadership (official or otherwise).
"It's just US, not EU."
Have you tried the 7b?
Employees of Bytedance (and other Chinese companies) have to deal with draconian employee rules and agreements that tie them back to the rules of their Chinese mainland parent corporation and the Chinese government itself. For example look at the details that came out in this lawsuit against TikTok, where employees have to agree to uphold Chinese national interests, uphold socialism, etc.
https://dailycaller.com/2025/01/14/tiktok-forced-staff-oaths...
Early TikTok was basically a standard Chinese startup in terms of how they collected data to optimize the app... It's not because they cared about what "you" are doing.
The Chinese tech ecosystem is just more competitive than the US one.
Most social media companies have been careless with data until they got public backlash though... TikTok's systems are probably more secure today than anyone else's... Because of the backlash.
WeChat is an app that works inside China, so obviously it's subject to Chinese laws.
All countries have some restrictions on what is and is not okay to have on online platforms. China just adds in "social disruption" alongside content that could promote violence (which would be blocked on Facebook, etc. also)
You need to not just drink in propaganda... When something just "can't be done" (because it is being monitored), then it just doesn't matter even if you have ill intent.
Does that sound like a business or an intelligence front?
Although, they do seem willing to sell potentially.
China restricted them from selling their recommendation algorithm, and it's hard to sell TikTok without.
But, TikTok did immediately start work on an independent algorithm that has no ties to China, so that they can sell it if it really comes down to it.
And you are telling me this is perfectly normal.
Tencent's Hunyuan 3D - Tripo, etc.
Tencent's Hunyuan Video - Sora, Runway, Pika, etc.
ByteDance's Trae - Cursor, etc.
DeepSeek R1 - GPT, etc.
Unitree - Figure, Tesla, Boston Dynamics, etc.
And a ton of this stuff is completely open source. All of the Tencent stuff is. This destroys the moat of so many companies that have spent hundreds of millions of dollars training and building their tech. It's just out there for free for anyone to build with.
And has anyone been checking the volume of Chinese AI research papers? Almost every impactful paper I've read in the last month has mostly Chinese names. And a lot of those papers come with code. Usually permissively licensed.
China is absolutely killing at the AI game. They've fast followed (or in many cases led) into a position of strength.
I wouldn't want to be RunwayML, Pika, Luma, or Tripo right now. Any "foundation model" company with a single use case is getting cloned and commoditized.
edit: Please don't downvote me because you don't like the message. I'm actually fine with you shooting the messenger, but this is absolutely worth talking about. It's pretty surreal to watch this all start to unfold over the last quarter or so. I want to read what others think about it.
It is that bad that all the American companies withrew from the public benchmark library. You can only see there Chinese commercial solvers and open source ones.
https://plato.asu.edu/bench.html
But in my opinion this was completely expected, most research in this area is now done in China, American universities have in comparison stopped in time.
Love this field of CS!
(I am not in AI, so please correct me if this is inaccurate.)
(Source: frustrated friends who lose the government contract to someone with inferior technology)
I just had to talk someone I know out of using RedNote as the TikTok shutdown loomed last week -- they had ZERO IDEA about its provenance. Unless you're paying attention, it's actually not clear what's what.
How many products keep sending all kind of unknown telemetry, have a ton of trackers, etc?
Can't really blame them doing the same.
1) They try similar like Zuck with LLama make sure GPT or Claude or Gemini is not the real winner in the West world like Android/iOS was for smartphone, Facebook/Whatsapp for social media/IM, Google Search for search, Windows for desktop OS. Majority people get used to to single product and they don't switch often.
2) Trying to kickstart community similar like LLama kickstarted big community around it that helped with tooling, testing, etc.
3) Slowing down development of ai companies in the west - giving them less data for training (after all midjourney, elevenlabs, llama kickstarted training on copyrighted content, it's pretty sure they using user data at least from e.g. free version of GPTo-mini) and bleeding their budget kind of war of attrition.
4) Familiarise the west audience with their ai models - after all still not many people using models like DeepkSeek since not many providers host them and majority of people don't have fast enough computer to run full models fast enough - Trae helps with that
Or are you being sarcastic? You must be being sarcastic.
If you really want an integrated experience, and not just a sidebar UI, you need to go the same route as Cursor and fork Code-OSS (the MIT-licensed part of VS Code, analogous to Chromium for Chrome)
I don't think it has any special api access?
Wondering if it is a fork of vscode though because right now it will be competing against cursor.
I've also blankly asked it if there's any issues with the current code, and found errors before I try to build / run a project.
But the moment it gets a little less basic, its a different story. I absolutely hate the experience. So I love it and I hate it at the same time. Is it going to replace me? I don't feel like its there yet. And this has been the case for a while now.
There is just too much money and people invested in this that its hard to say anything negative about it. Nearly every VC has rebranded their websites around an AI-driven future. And, to be fair, it’s not a fad—it genuinely works well for many things. But for now, I’m still skeptical about how far it can really go.
I'm terrified to install a binary like that
https://en.m.wikipedia.org/wiki/Metalhead_(Black_Mirror)
Use something like emacs/vim/kate/eclipse/qtcreator if you want to avoid "phone home" software.
If a Chinese company does the same, at most someone in China will shrug.
Does anybody feel the same? I feel like nobody touched upon this, but it's always very nice to feel interfaces are responsive.
I accessed this through a Qubes AppVM (no GPU, limited memory and CPU budget) and the presence of videos makes this a very slow scrolling experience for me.
In general, anything that involves JS/CSS animation/blur/effects makes sites pretty slow (up to unusable for me). The unlogged homepage for github.com for example, spins my cpu at 100%.
But I guess instead of writing great libraries to avoid repeating ourselves, we drop all code to a lake monster which eats it and spits out answers.
Thus it supporting English and Chinese languages.
And it may be a competitive step taken after Alibaba launched their new coding assistant.
Although I haven't tried to find a Chinese mode myself.
That it's targeted at Chinese abroad is just what I saw from Googling it and reading the SCMP (Singapore) article on it.
The company that made it is also from Singapore (even if it's under Bytedance's umbrella).
They seem to be separate from TikTok. The only thing they have in the app store for example, is some chatbot application.
¹: which is dozens of them at this point, open and not.
²: with lots of speakers and training data.
Separately, the fact that they can translate between languages where they have never seen any translation data in their inputs shows that they have internal models of the world and language, that go beyond what one might expect from a statistical parrot.
They do have world understanding - perhaps limited some by the fact that their input data may not cover a lot of everyday things.
Regardless, what is this exactly?
Also, anybody knows a decent IDE where AI is a first class citizen? So far my experience with plugins is, that they have too little context to be actually useful.
Sidenote: That marketing video was nauseating, it was moving too fast and didn't show any features for long enough or for enough steps.
If I ever have a need for buggy crap code, any random boot camp dev can write it for no money at all.
Makes you think about how they're extracting value from you to make up for this. I rack up non-significant amounts when using Claude 3.5 Sonnet for coding.
Considering the data they siphon from their other "free" apps, I don't want to think about what this does on developer machines with code bases and production access.
Recommended.
(But of course it would be nice to be able to run it against a local model which doesn't seem possible at the moment.)
Zed only has syntax highlighting and an ai assistant, but that's it. Even though Vscode (and the rest) is still also considered a text editor and not a IDE, they have by far way more features.
Sure, you get super duper high speed, but at what cost?
- Void
- Melty
- Pear
- Aide
Just another AI wrapper around VS Code smh
Effort would have been much better spent on getting AI auto-complete "merged" with LSP. But who is listening? Pretty much everyone these days think they can plug an LLM and have their life issues sorted out.
It's sad because I have seen some truly remarkable progress in LLMs, but I feel like we aren't allowed to be honest anymore that LLMs aren't going to replace programmers or moderate our expectations.
For any semi novel development I want to write it myself to establish the patterns I like and get an understanding of what I'm building. It's the boring stuff I hand over to Claude.
I certainly can see it being useful for writing additional tests, although I am very worried based on what I've seen that it will introduce more bugs than it's worth, which is why it's still a tool I use for hobby projects and not for real work.
If your metric is number of tests, sure. If your metric is number of useful tests… eh
But literally every single programmer I know has this opinion.
Of course they are terrible at their job, so AI doesn't make their code much worse.
https://code.visualstudio.com/docs/copilot/copilot-edits
I've done some stuff for a compiler in Java, backend web services in Python/Go/TS, frontend w/ React/TS.
I was a huge skeptic of this stuff at first but started using it about a year ago. I could definitely live without it, but it also saves me a significant amount of time.
I've been working on a feature in Python. With Copilot edits I just needed to find files the implemented the pattern, add it to the chat context. and write something like "implement feature x following the same pattern". It never gets it right the first time, but you can just keep the conversation going and have it iterate.
Afterwards I can just write /tests and Copilot generates reasonable tests. If it missed cases I can ask it to write cover those tests. Often times I can also just write literally "cover edge cases" and it handles all reasonable scenarios.
With what I was left with at that point, it probably would have been a wash to fix that code or just start from scratch manually.
I don't blame Loveable, I think it's a wonderful product bordering on magic, like v0 and Bolt.new. I just think they are amazing products that have been over hyped to god-like status.
What it is just a summary of the internet.
I used to think he was competent but now I think he's a moron
[1] https://arxiv.org/abs/2404.03502
I am not a big fan of LLMs so I try them once in a while, asking to "implement blocked clause decomposition in Haskell." They can recite a paper (and several other papers, with references) almost verbatim, but they do not posess enough comprehension of what is going on in these papers. As time passes by, with each new LLM, the level of comprehension drops. Last time LLM tried to persuade me to write code instead of providing me with code.
With this example usage:
How did it do?Did it do just from the prompt or you had to nudge it? Can you share full chat history?
Can you share actual interaction on their site?
https://support.anthropic.com/en/articles/9450526-how-can-i-...
Of course, I believe you. It is verbatim.The training data is always limited, because if there was an existing software that already did what I'm doing, I'd be using it instead of writing it :)
So by definition, unless I'm learning how to program and doing exercises that thousands have done before me, I'm doing something for which there is no training data.
Doing basic stuff with a very limited scope that’s likely well documented all over the internet, great.
Boilerplate, tedious yet simple things, works awesome as enhanced autocomplete.
More complex stuff, give it a shot, maybe you’ll get lucky, otherwise if it starts fucking up I’ve had the most success just taking a step back and doing it myself, maybe chatting with it as a live docs substitute, or a realtime stack overflow/discord programming channel which are also sometimes of dubious quality but frequently useful.
The misery really lies in getting stuck in that cycle of it just messing up over and over as you try to get it to make this thing work that is clearly beyond its scope and it’s just turning everything into a greater and greater mess of hallucinated bullshit.
Yeah, I think this is exactly right. I use it for adding new files to an existing codebase, but without giving it access to that codebase, by passing in the definitions I want it to work with. It gets stuff wrong a lot, and I need to keep anything I'm asking for _well_ within a scope that I can be eagle-eyed about. But if I want to do something pretty simple that would be slightly annoying to write by hand, and can be done in a single file, it makes a pleasant alternative to typing out the code by hand.
I don't think I would be able to do this as a junior developer, I think this is only working because I can tell when it's full of shit, and I lasted about 90m of being willing to let IDE-integration happen, because it's too easy to be lazy and not scrutinize every little change, and that way lies total madness. This makes me beyond skeptical of non-developers writing anything significant with it: they'd be much, much better off learning Bubble or similar.
I've built a couple pretty simple apps that have been basically no code beyond tweaks for me.
I do think a big knowledge gap is the acceleration of prompting by knowing how to talk about code. LLMs show a lot of difference in response between "make the top bit stay on the screen on small screens" and "make the <th> element sticky for viewports <600px".
- autocompletion (about half the time)
- questions on the syntax for certain commands (faster than looking up in the docs, usually pretty accurate)
- questions for how to solve a particular problem or gotcha (faster than looking up on SO, but also less accurate)
It's a useful tool for certain cases, for sure. But it's not a "game changer" in terms of productivity.
So I definitely think the value proposition of AI coding is higher when you have programmatical control over the workflow.
On a related note, the recent Cline Plan+Act function has also been a game changer.
LLMs are a tool. It’s also weird and counterintuitive so it takes some time to get really good with them.
I haven't heard anyone in tech say programmers will be replaced by this tech.
I have heard persons outside of tech say it, but I think they lack sufficient context.
I mean, if we get AGI, yes, we might be fully out of a job. But, so far I think many programmers agree this seems to be a ways off.
But, I think you underestimate how many programmers used to copy paste code from StackOverflow and articles, etc.
We even had/have a phrase for that: "copy pasta".
Those programmers are getting more applicable templated code than they used to get via copy paste.
When I'm coding in my preferred languages, I am faster without AI.
But, when I'm writing yaml or something else for an unfamiliar tool or platform, I do get a productivity boost, even if I have to debug the code / configuration.
Also don't believe your lying eyes.
Everything that has been coming out of China is better than anything our mega trillion dollar corporation are producing and they are doing with with sanctions on the best GPUs.
I don't work in big tech, why are they so bad with sigifically more money and more advanced technology?
Google had a decade and a half head start and they can't even produce a model that is as good and as cheap as Deepseek building with no compute compared to Google.
I have been using Trae and VS Code with Copilot is laughable compare too it despite using the same models. So we can't even prompt LLMs as good as they do and they don't even speak English as a first language LMFAO.
We might be fucked and this is just the beginning.
Maybe time to learn code should be time to learn Chinese.
At the forefront of our politicians’ minds are how many genders there are. In the UK just building a new high speed railway has practically been the Manhattan project. Will we get back on track? I think the chances of that are rapidly diminishing.
This seems so irrelevant. Why do you think reactionary politics will lead to more innovation?
Only problem is now I have no idea what they are talking about half the time.
The Builder tool is probably the best feature of Trae. It builds React UIs better than any tool even v0 which is built for building React UIs. It very good at recreating UI from image even better than using the model directly which I'm not too sure how...
Code completion is probably closer to copilot than cursor so it depends on how hardcore you like your AI code completion. So Cursor might be better especially for repetitive refactoring.
The overall design and feel of Trae is better the VS Code and Cursor, but I'm a big fan of JetBrains and it feel a lot like a JetBrains IDE.
It's basically a better Fleet. In fact, it's if Fleet and VS Code had a baby lol.
Like why does this page even exist? https://www.jetbrains.com/ides/#choose-your-ide
https://www.jetbrains.com/fleet/
US firms might have the lead in terms of absolute best performance, true. But their prices are 10-20x more for a product that is marginally better at most. When it comes to coding DeepSeek V3 is very close to Claude 3.5 Sonnet and exceeds GPT-4o.
Some C-levels newphew into computers and need a school project or something?