I think the solution here would be to write a hand-written letter.
Sure, someone can make AI write a letter with some kind of contraption holding a pen (I think StuffMadeHere did something adjacent to this). But it would likely be more obvious, plus it requires physical actions and a stamp. All things that low-effort AI spammers aren’t going to bother with.
Physical letters do not obviate scams, nor is the cost that prohibitive. I remember actual 419 scams on blue airmail all-in-one letters back in the 80s. And that was international post too.
I have an inbox, and I do not receive a lot of scam post. In fact, I don't think I received any since I lived at this address (~10 years ). We do get a few promotional leaflets every other week.
OTOH, I get hundred of spam emails every day.
The former is something which I can handle manually easily, the other is not.
If you are targetting a list of well-known authors I guess outsourcing the writing of a couple of hundred handwritten letters shouldn't be too hard. I'm sure they they can find a school class in Nigeria or Kenya who would gladly do it for a few dollars — or a struggling teacher willing to get creative with the homework assignments.
Uh - there are entire political lobbyist organizations that use something similar to an "autopen" to make mass letters personalized and appear to be handwritten...
Heck - I have seen some in the mail from the "sell your house for cash" companies - typically behind the friendly, "homespun" personable facade, it is a REIT (real-estate investment trust) - or something similar...
(Myself, I can tell that these are mass-generated - but I am (at least also at this point in life - who knows when I get much older) easily able to tell a scam email, phone call or txt-type message - I can typically spot the signs - but those signs are typically there to "weed-out" the people that won't fall for the scam anyways...) - but my non-cynical, non-technical, non-paranoid friends and family need assistance spotting these...)
> If you’re a scammer who uses “AI” to try to defraud actual humans, please die in a fucking fire, thanks.
Refreshingly direct and unfiltered, despite Scalzi being a well-established writer.
If you are looking for a refreshingly fun light read to brighten up your day¹, try Scalzi's When the Moon Hits Your Eye (2025), in which the moon turns into actual cheese.
1: It includes the horrific death of a Musk/Bezos-like tech-bro with more money and tech than sense. Good fun!
More and more of the internet of humans need to rely on recommendations of other humans. Lobste.rs and other like such that retain the tree of joined people could work for other communities as well. Kind of like return of the FTP warez scene of 90s but for the rest of us.
AI can trivially replicate average-ish and somewhat above human writing if given a tone sample to copy. Getting to replicate the quality of writing output of a decent author, probably not without a lot of effort, but the threshold here is to sound like a plausible e-mail from a book club or similar group, not to write the Great American Novel.
Categorising all the emails is not the challenge. The author needs to extract a very small number of genuine requests from the slush pile. which is a different problem, and much easier to set expectations in the community for.
I can completely understand not wanting to deal with the hassle, but pretending it's all about AI is disingenuous in my view.
The author needs to avoid a sufficient number of false positives for the time investment not to be prohibitive, and that is what he is arguing is becoming a hard problem. I have problems believing that given some of the e-mail I receive. I have no problem trusting Scalzi on this.
In my real world experience it's easy to fix. Most spam is generic. Publish a blog post asking applicants to include a specific keyword in the subject line. That sorts out 80% of the spam. (probably all of it)
Asking for a cover letter in docx format, requesting info on the format of the book group, and what other authors they have discussed recently, sorts out another 99%.
Filter both these out and you are left with a small number of applicants. If someone is a tailoring an AI to defeat this, then author has a very high value event on his hands that he should hire someone to help organise.
If applicants are not willing to do this, then they clearly are not offering a high-value opportunity in the first place. His excuse obviously fools most people, hence your reply, but it's very unlikely to be the big picture in my view. He just doesn't want to do the book group. Not enough to set up some simple filters anyway.
Or you have someone running an AI bot that does research on your target automatically. Ironically, one of regular features of spam I'm getting semi-regularly now is for "marketing" services" that provide OpenClaw instances to research and individualise messaging.
he is not saying he doesn't want to do bookclubs he is saying he does not want to spend more time to do bookclubs than he previously did, it is not beyond the wit of man to efficiently pick out a genuine request, but even with efficiency time and effort required is not null.
If there is more slush and the fraudulent slush increases in quality and scope it must mean that the wit, effort and time required must also increase to deal with the problem.
Dealing with problems that do not give one a return on the investment seems not worth doing, especially when it turns out the most efficient way to deal with this particular problem is to absolutely refuse to do any book club stuff.
generic emails sure, but harder to conjure up a convincing picture of a specific book club, where it is, who will be there.
If people are taking the time to generate this kind of AI invite, then it must be a very high value event. Possible, but I suspect there are more mundane reasons for avoiding the admin
There are plenty of examples of AI being successfully used to emulated the email / messaging style of a specific individual already known to the target, for spear fishing attacks, and fake video and audio of family members tricking people. I think you're substantially underestimating the peak ability of AI these days
I'm not saying AI is incapable of these attacks, I'm arguing a more likely explanation exists. If he wanted to accept, say, one book club a week, I don't believe he would have too much trouble figuring out a way to safely receive applications
a lot of people , including myself, are using AI as an excuse to push thru awkward changes
I'm not doubting AI spam is an issue, but to solicit one book club appointment a month, solutions exist. It wouldn't be hard to identify the most genuine invites. Even if the middle ground is increasingly hard to filter
Sorry, why do we care about this one dude's opinion?
I host book clubs and we always have a fantastic time.
Although my writing style/unhinged nature makes it pretty obvious no LLM would ever write like this. Hedonist Philosophers are not exactly what LLMs were trained on.
Sure, someone can make AI write a letter with some kind of contraption holding a pen (I think StuffMadeHere did something adjacent to this). But it would likely be more obvious, plus it requires physical actions and a stamp. All things that low-effort AI spammers aren’t going to bother with.
Contact me: letter > envelope > stamp > post box
I have an inbox, and I do not receive a lot of scam post. In fact, I don't think I received any since I lived at this address (~10 years ). We do get a few promotional leaflets every other week.
OTOH, I get hundred of spam emails every day.
The former is something which I can handle manually easily, the other is not.
Heck - I have seen some in the mail from the "sell your house for cash" companies - typically behind the friendly, "homespun" personable facade, it is a REIT (real-estate investment trust) - or something similar...
(Myself, I can tell that these are mass-generated - but I am (at least also at this point in life - who knows when I get much older) easily able to tell a scam email, phone call or txt-type message - I can typically spot the signs - but those signs are typically there to "weed-out" the people that won't fall for the scam anyways...) - but my non-cynical, non-technical, non-paranoid friends and family need assistance spotting these...)
They're "physical letters but digital," tied to a human identity and with proper proof of receipt.
Refreshingly direct and unfiltered, despite Scalzi being a well-established writer.
If you are looking for a refreshingly fun light read to brighten up your day¹, try Scalzi's When the Moon Hits Your Eye (2025), in which the moon turns into actual cheese.
1: It includes the horrific death of a Musk/Bezos-like tech-bro with more money and tech than sense. Good fun!
To be clear, if he wanted to accept a book club invite every month or so, that would be quite easy to achieve. I doubt AI is the issue here
'Bluntly, I can spend my days sorting “book club” spam, or I can write books. One pays me money. The other does not. '
erm, doing the actual book club doesn't pay either and is going to take a lot more energy than selecting a genuine invite from the slush pile.
I can completely understand not wanting to deal with the hassle, but pretending it's all about AI is disingenuous in my view.
Asking for a cover letter in docx format, requesting info on the format of the book group, and what other authors they have discussed recently, sorts out another 99%.
Filter both these out and you are left with a small number of applicants. If someone is a tailoring an AI to defeat this, then author has a very high value event on his hands that he should hire someone to help organise.
If applicants are not willing to do this, then they clearly are not offering a high-value opportunity in the first place. His excuse obviously fools most people, hence your reply, but it's very unlikely to be the big picture in my view. He just doesn't want to do the book group. Not enough to set up some simple filters anyway.
If this guy wanted to do bookclubs, he absolutely could. It's not beyond the wit of man to efficiently pick out a genuine request from the slush pile
If there is more slush and the fraudulent slush increases in quality and scope it must mean that the wit, effort and time required must also increase to deal with the problem.
Dealing with problems that do not give one a return on the investment seems not worth doing, especially when it turns out the most efficient way to deal with this particular problem is to absolutely refuse to do any book club stuff.
It can definitely replicate a human-written email.
If people are taking the time to generate this kind of AI invite, then it must be a very high value event. Possible, but I suspect there are more mundane reasons for avoiding the admin
a lot of people , including myself, are using AI as an excuse to push thru awkward changes
I know a scapegoat when I see it
I host book clubs and we always have a fantastic time.
Although my writing style/unhinged nature makes it pretty obvious no LLM would ever write like this. Hedonist Philosophers are not exactly what LLMs were trained on.
We also care about it because it's an indicator of the rise of a new societal problem of signal being even further drowned out.