I got the pop-up and had a little read and it's framed as an "Interview", presumably by some AI interviewer they've set up, and I immediately realised they're probably aiming for a two-fer, test out their AI interview thing, and slurp up ideas from users.
I draw the line at paying them for the "service" they're "providing".
They'll be scanning this thread as well of course, I won't participate in some user study without compensation, and will feel double insulted when they don't even consider utilizing human time to conduct it.
I'd never wanna signal that AI interviews are at all acceptable. The bar should be equal amounts of human consideration on both sides, otherwise I would feel its likely disrespect to the value of the human forced to interact with AI instead. I'd also venture to guess that these tools are NOT good enough to pick up on anything besides transcripts, which leaves a TON of actual human context on the table.
I can see a world where they get better at that visual and cadence based context, but i just know its not there now, and these interviews are likely on some absurdly subpar model that is not close to up to that task.
In all likelihood its as painfully pathetic as ai transcribes the audio to text, and interprets the text without anything extra, then responds(not to say that it isn't impressive, its just largely worthless in exchange for human on human interviews).
> Gain ammo for regulatory capture from anti-ai sentiment, check.
I’m not sure what you mean by “anti-AI sentiment” here, because what I would interpret that as (and generally agree with myself) would answer the title question “What do you want from AI?” with “for it to all be agreed a bad idea, deleted, and made illegal”.
This kind of stuff is cute and probably well-meaning when it's coming from an underdog. It's not cute nor well-meaning when it's coming from a $2T company.
I have no doubt that they are already spending tens of millions on precisely understanding the sentiment of users and non-users, in ways that are precisely targeted, in sessions that occur in controlled environments. They're probably also sentiment-mining their chat logs and the internet at large. Whatever the purpose of this study, it's almost certainly not what it says. I suspect one of the following three objectives:
1. Gain more diehard followers by projecting the image of a company that "listens".
2. Ask loaded questions to convince you of their beliefs.
3. Score cheap talking points for regulators, such as that in their online survey, 90% of people love / fear AI.
Can't they just use AI to generate those answers? Why do they still need humans for this? Surely AI is advanced enough to not need human input anymore. We have AGI as promised right?
Today's AI is a lot closer to how dumb me was imagining computers back in 1985, when I had my first interaction with a Romanian clone of ZX-Spectrum in the school's newly equipped computer science laboratory: https://retroit.ro/wp-content/uploads/2019/12/IMG-20200707-2...
Must say, I was quite afraid to even press the keys at the time! My brothers, many years later, had an epiphany on why that was. You see, I was a countryside boy. I grew in a world where becoming a tractor driver was a sought of career. Thing is, a tractor is a mechanical contraption. And you know what mechanical contraptions have in common? You can't just pull levers and push pedals at random without knowing what you're doing and expect the thing to work, au contraire! You'll break the expensive damn thing.
So my first interaction with the KEYBOARD of the Home Computer '85 was overly gentle - which probably helped because they were shit.
But beyond that, nobody told me what computers are good for and I could only imagine they could help me solve the numerous math problems that were left as "exercise" in the textbooks that gave two helpful examples "1+1 = 2" and "1-1=0" and then posted the problem without the answer: "now prove the Fermat theorem!"
Fuck math and fuck the educational system! The only purpose of those textbooks, just like AI, is to provide me with WORKING examples from which I can LEARN. That's the whole point of living, learning to solve increasingly harder things so eventually YOU can be the frontier. And no AI will take that from me, they are just glorified textbooks.
Money is not a resource that gets used up when you spend it. It's a virtual "I owe you" that just changes hands. Nobody incinerated billions, some people spent billions, other people earned billions from this transaction. You may not like who earned the money, but we kinda agree as a society about that it was not yours to decide.
I answered this last night, though, I wanted to share a lot more with them, but I don't like giving ideas for free. There's so many things these companies can be doing that I keep to myself but I'm mostly convinced a lot of them need to take a step back and rethink LLMs, LLM design, etc because it looks to me like we're all stuck forcing ourselves to use an insane amount of compute, with an insane amount of data, but the small models have shown to be capable of reasonable reasoning on way less resources and data. Heck, the distilled or quantized version of the only open model OpenAI ever released is insanely good.
I don't know all the answers, but I think we're seeing the "ceiling" of where LLMs can go, and we're trying to force them to go further in the hops of achieving AGI somehow, but AGI in my eyes is not going to work with the current design for LLMs unless they're actually building AGI completely differently similar to what I assume it could / should look like, then I'll give them that, but none of them have shown any signs of true AGI. Sure an LLM can be impressive in terms of speed of outright (and correctness now!) but its not AGI. AGI isn't a buzzword you change, it was defined before Sam Altman and Dario tried to redefine and water it down for IPO gout.
I agree entirely. The problem is these companies and the American public find themselves in a tough situation where saying “maybe we don’t need to build out this insane amount of infrastructure” causes everything to fall apart for both the companies and our economy.
Don’t get me wrong, I 100% agree that we should do it anyway instead of postponing it. Still, that’s why it won’t happen.
I think the real wake up call will be as we start focusing more on smaller local models, because a lot of industries will not touch AI with a 100000 foot pole, but I could see them warming up to something like Google's Gemma, or Microsoft's Phi, 100% local, on a corporate system.
I got the popup and clicked away because I don't have time or tolérance for AI interactions, but a friend of mine has been organizing people with critical opinions to give their takes as a way of putting strong criticisms of AI companies on their own website and reach out to the anthropic workers to try and make them see they are building the panopticon, I'm skeptical that anthropic won't censor everything but would be interesting to see whether that effort will actually move the Overton window a bit
I must be one of the few who completed the interview. I was hesitant with some questions but still gave direct feedback and asked to make my feedback public. Not sure if they will do it. And just like LLM does the interview did pick up some nuances from my answer and played them back for me to confirm. "Oh so you are saying not only this but that, is that right?
Whether they use it or not, whether they use it to manipulate us or whatever, it is undeniable that AI is the biggest change happening right now, and if there is an opportunity to nudge it, why not.
Edit. And no, I was not asked to login. They likely had the login. I just clicked the survey link from the terminal claude, it opened in a browser and off to the races.
This is, of course, marketing. If Anthropic wanted to be a good sport and learn from users, it could just ask Claude to grep user conversations and support chats for the f-word.
AI needs to be open weights at a minimum, but ideally open source. I'm not terribly worried as it is strictly inevitable. But having these companies steal work from humanity and use it to train their models while attempting a hostile takeover of the economy is morally bankrupt. I lose a lot of respect for anyone going to work for these labs.
Surprising number of comments in here complaining about a company asking their users for ideas, when that's pretty much the first piece of business advice any startup will get.
Surveys typically ask people multiple-choice questions, but not what they meant by their answers. This interviewing tool apparently allows freeform input? It might be interesting if a proper survey combined the two, so you could see what people picked and also what they meant by it.
I responded. The agent happily "nodded along" as I explained how I haven't yet seen any real great benefits from ai. Sure there are some things but always at the cost of something that i think matters more. I use LLMs daily fwiw.
- AI is the largest financial fraud since Madoff. Money is printed to run companies at a deficit. Then these companies IPO with an artificially small float to prop up the valuation. Then the fake $2 trillion valuation can be used for the fraudsters to borrow money against the fake stock, which creates even more money without any useful service being provided. The public pays in inflation.
- AI has turned the Internet into the Slopnet. More and more semi-serious websites look like SEO websites.
- AI has destroyed large parts of programming, where subservient engineers implement Rube Goldberg production pipelines solely to please management. No software or SaaS has gotten better since 2022.
- AI dominates the discussion space, and clear thinking people are forced to fight back against the advertising and the insanity.
Thank you for your load-bearing enquete and I hope the seams are watertight!
I just ordered a couple hundred "stop the slop" stickers to stick onto obviously AI generated posters, ads etc in public, I doubt anthropic is genuinely interested in my opinion
I’ve seen this so many times before, in startups, and in enterprises. I want to be charitable to these feedback efforts, but my hot take is “we’re out of ideas, help please.” Always welcome and hopeful to be proven wrong.
I find anthropic to be such an unsavoury company. I am not an AI doomer and frankly I don't care much about datacentres and so on, it's just how scammy they come across.
The authors of Claude Code churn out so much BS that gets just eaten up by credulous engineers & engineering leaders. "I posted 250 changes today with claude code!" - author of claude code. What's your point? If we don't care about quality, I can output 50,000 changes in a day.
I see them as 95% sales 5% engineering and it's irritating to have marketers constantly throw gasoline on the AI hype fire & have people still regard them as engineers. The claude models themselves are alright.
I started this thinking it was going to be an actual interesting interview process, but it felt like a typeform page once it get going. Had it been a typeform page I might have completed, but framing it as an interview annoyed me for some reason and I quit after the 4th question.
Same experience. The question "what is your most memorable experience" also kind of annoyed me, because I see AI mostly as a tool. It is like asking me what my most memorable experience with my iphone is. I'm sure I could cook up something, but it felt like they were mining users for happy PR.
I think it's a no-win situation, because I also think of that talk. But if I credited the saying to Monterio, I'm sure someone would remind me that he sourced it from Goodfellas.
I draw the line at paying them for the "service" they're "providing".
I'd never wanna signal that AI interviews are at all acceptable. The bar should be equal amounts of human consideration on both sides, otherwise I would feel its likely disrespect to the value of the human forced to interact with AI instead. I'd also venture to guess that these tools are NOT good enough to pick up on anything besides transcripts, which leaves a TON of actual human context on the table.
I can see a world where they get better at that visual and cadence based context, but i just know its not there now, and these interviews are likely on some absurdly subpar model that is not close to up to that task.
In all likelihood its as painfully pathetic as ai transcribes the audio to text, and interprets the text without anything extra, then responds(not to say that it isn't impressive, its just largely worthless in exchange for human on human interviews).
Gain ammo for regulatory capture from anti-ai sentiment, check.
Market research from pro ai people, check.
A bunch of new signups to fuel user numbers, check.
Advertising for their new survey product, check.
A bunch of new stolen training data, check.
Me participating in this scam? Uncheck.
I’m not sure what you mean by “anti-AI sentiment” here, because what I would interpret that as (and generally agree with myself) would answer the title question “What do you want from AI?” with “for it to all be agreed a bad idea, deleted, and made illegal”.
"stop building ai data centers because of the environmental damage" -> "we need to pace the frontier"
"don't build AGI and doom us all" -> "we need to pace the frontier"
"stop shoving ai into everything" -> "we need to pace the frontier"
"eat poo dario" -> "we need to pace the frontier"
I have no doubt that they are already spending tens of millions on precisely understanding the sentiment of users and non-users, in ways that are precisely targeted, in sessions that occur in controlled environments. They're probably also sentiment-mining their chat logs and the internet at large. Whatever the purpose of this study, it's almost certainly not what it says. I suspect one of the following three objectives:
1. Gain more diehard followers by projecting the image of a company that "listens".
2. Ask loaded questions to convince you of their beliefs.
3. Score cheap talking points for regulators, such as that in their online survey, 90% of people love / fear AI.
Not only Star Trek were visionaries!
Today's AI is a lot closer to how dumb me was imagining computers back in 1985, when I had my first interaction with a Romanian clone of ZX-Spectrum in the school's newly equipped computer science laboratory: https://retroit.ro/wp-content/uploads/2019/12/IMG-20200707-2...
Must say, I was quite afraid to even press the keys at the time! My brothers, many years later, had an epiphany on why that was. You see, I was a countryside boy. I grew in a world where becoming a tractor driver was a sought of career. Thing is, a tractor is a mechanical contraption. And you know what mechanical contraptions have in common? You can't just pull levers and push pedals at random without knowing what you're doing and expect the thing to work, au contraire! You'll break the expensive damn thing.
So my first interaction with the KEYBOARD of the Home Computer '85 was overly gentle - which probably helped because they were shit.
But beyond that, nobody told me what computers are good for and I could only imagine they could help me solve the numerous math problems that were left as "exercise" in the textbooks that gave two helpful examples "1+1 = 2" and "1-1=0" and then posted the problem without the answer: "now prove the Fermat theorem!"
Fuck math and fuck the educational system! The only purpose of those textbooks, just like AI, is to provide me with WORKING examples from which I can LEARN. That's the whole point of living, learning to solve increasingly harder things so eventually YOU can be the frontier. And no AI will take that from me, they are just glorified textbooks.
I don't know all the answers, but I think we're seeing the "ceiling" of where LLMs can go, and we're trying to force them to go further in the hops of achieving AGI somehow, but AGI in my eyes is not going to work with the current design for LLMs unless they're actually building AGI completely differently similar to what I assume it could / should look like, then I'll give them that, but none of them have shown any signs of true AGI. Sure an LLM can be impressive in terms of speed of outright (and correctness now!) but its not AGI. AGI isn't a buzzword you change, it was defined before Sam Altman and Dario tried to redefine and water it down for IPO gout.
Don’t get me wrong, I 100% agree that we should do it anyway instead of postponing it. Still, that’s why it won’t happen.
Whether they use it or not, whether they use it to manipulate us or whatever, it is undeniable that AI is the biggest change happening right now, and if there is an opportunity to nudge it, why not.
Edit. And no, I was not asked to login. They likely had the login. I just clicked the survey link from the terminal claude, it opened in a browser and off to the races.
No, you're supposed to pay them to take your job. And be happy about it!
Why are you doing life wrong?
Talk to your users!
I'm looking forward to seeing the results.
- AI is the largest financial fraud since Madoff. Money is printed to run companies at a deficit. Then these companies IPO with an artificially small float to prop up the valuation. Then the fake $2 trillion valuation can be used for the fraudsters to borrow money against the fake stock, which creates even more money without any useful service being provided. The public pays in inflation.
- AI is a bubble where companies hoard unconnected GPUs, which drives up prices of GPUs and RAM: https://www.wheresyoured.at/wherere-all-the-ai-chips/
- AI has turned the Internet into the Slopnet. More and more semi-serious websites look like SEO websites.
- AI has destroyed large parts of programming, where subservient engineers implement Rube Goldberg production pipelines solely to please management. No software or SaaS has gotten better since 2022.
- AI dominates the discussion space, and clear thinking people are forced to fight back against the advertising and the insanity.
Thank you for your load-bearing enquete and I hope the seams are watertight!
… and don’t vandalize other people’s work just because you think it might be “slop”.
The authors of Claude Code churn out so much BS that gets just eaten up by credulous engineers & engineering leaders. "I posted 250 changes today with claude code!" - author of claude code. What's your point? If we don't care about quality, I can output 50,000 changes in a day.
I see them as 95% sales 5% engineering and it's irritating to have marketers constantly throw gasoline on the AI hype fire & have people still regard them as engineers. The claude models themselves are alright.
They are not interested in my thought though. They just want attention.
Good quote, though. And good philosophy.