You should add a parallel timeline of how many times AI CEOs have claimed their next model is too dangerous to release, AGI is months away, or some white collar job will be obsolete within 6 months.
I don’t know anyone in the tech industry who thinks AGI will never happen or that software engineering and white collar jobs can’t be automated. We all read sci fi, you’re not some unique visionary for anticipating AI. The frustration is with how much the claims have outpaced reality and how poorly the investors and executives have treated their workers during this transition.
Yeah reading this timeline wasn't satisfying to me for exactly this reason: It didn't give the counterarguments that were being made at the time of each "goalpost".
My (probably flawed, but still) memory is that the first one of these threads I participated in, at the end of 2022, was saying that none of us would have jobs after two more years of development of these models. Two years from then was almost two years ago now, and we're still "a year or two" out.
On the flip side, the thesis that these will be useful tools that will augment the work of software developers when understood and used for the things they are good at has (IMO) remained undefeated during this entire period.
I’ll go on record and say that AGI will never happen (in the next 50 years). I think that’s also the timeline for white-collar job automation that requires critical thinking.
But the author acknowledged times that someone said something stupid about AI?
He's not claiming there's no flaws or that there aren't CEOs claiming more than is possible right now. He's making the point that these don't negate the parts that are genuinely impressive.
I've encountered waaaay too many "you can't possibly have done that with AI because AI occasionally hallucinates" and "CEOs say things for marketing therefore AI can't really do anything" type of posts.
Understandably, the frustration is also around "what happens then?"
Everyone is wondering about what happens if (when?) it finally comes true
We really have to figure out what comes next for your average person. I think the reason no one wants to talk about this is because the answers aren't great. The average person not going to be living a great life in all likelihood, once we have no access to capital anymore
Yup, that's the weird thing about the booster argument. "We finally achieved it, we're no longer important or the dominant species and Sam Altman and Dario Amodei run what's left of the world from their bunker!" Yay?
I wouldn't be surprised if multiple nation-states already had them without calling them such because 'Look, someone pushed a button 3 hours ago before it engaged'. Once the private capital gets their hands on them, we immediately transition to cyberpunk no matter what else happens in the world.
The average person will not have any life in all likelihood. That is the end goal of capitalism. However, shortly after everyone dies, the remaining survivors - the likes of Musk, Thiel etc - will find themselves roughly equal and needing to establish a new hierarchy.
You don’t know me, but I am in the tech industry and believe AGI will never happen.
Albeit that is because I did a BSc in psychology where I developed a deep distrust for intelligence research and concluded that intelligence is not a useful term in philosophy nor science (and especially not in engineering).
I strongly agree. In scientific terms, we already know how to achieve 1G constant acceleration space flight. "All we need to do" from an engineering standpoint to achieve it is develop miniaturized fusion reactors and superconducting electronics. Hell, something approximately as good was achievable since the 1950s with a gargantuan ship that pooped out atomic bombs through a giant shock absorbing pusher plate[0]. Trivial, right? Except it isn't. Both of these projects, which are scientifically feasible, are completely impossible to actually build. They are off the table in engineering terms.
We are so much closer to 1G constant acceleration space flight than we are to AGI. We know, in principle, how to achieve 1G travel. We don't know, in principle how to achieve AGI. Our best guess so far is something along the lines of "emergence" which means "maybe if we do enough matrix math in the right way it'll wake up and become a being with agency and intelligence". Another way to say this is "hopes, prayers, and lots of GPUs".
Let's all get a grip. Without a coherent theory of intelligence, you aren't gonna make one in a lab. That's not how science works, it's not how engineering works. Start at the beginning.
ELIZA could pass the Turing Test and is over 50 years older than GPT2. You are claiming we have had AGI available since before the first version of Unix came out?
Literally every CEO AI or otherwise claims bullshit about their product. It's their job. It's got nothing to do with people actually using something and not coming to terms with it having any value. If you've never seen a screwdriver it's a waste of time compared to nails.
“Everyone else is doing it” is a childish justification and there’s a line where this becomes fraud and a felony offense. Do you think Elizabeth Holmes should have been allowed to make up claims about their blood testing technology? Is it OK to grossly overestimate Claude’s capabilities when the US military starts using it to determine strike targets?
I'm not justifying anything nor am I arguing against it. I'm saying it's not a thing worth mentioning as it's not related to AI in any way. If one would try to find AI related negatives go for datacentres, chip shortage/price hikes etc... So many better arguments to choose than just saying "CEO's are lying!!1!".
It's their job to lie and defraud the public? Why don't you believe there shouldn't be consequences for lying and defrauding the public? Do you believe tech CEOs are above society and decency? Does their wealth make them immune from accountability?
It's gotten so bad that I feel that the majority of HN blog posts on the topic of AI are partially or fully AI-generated. Even the ones that are highly critical of the AI industry. People have just spent so much time next to AI, soaking it in, that they're becoming dependent on it - either because they actually can't write for themselves anymore, or because they care about a topic so little that they can only squeeze out a short paragraph and then ask their LLM to fill in the blanks for what were supposed to be their thoughts. For AI boosters it's somehow even worse.
To be fair I think we'd be able to claim AGI is here if that problem is solved. At this point the models are so smart they're borderline super intelligent if they were cognizant of hallucinations and their own shortcomings. If GPT 5.5 or Opus 4.8 could tell you "I don't know" they'd certainly be "smarter" than any individual human. Some specialists might be better in niche domains, but I don't know of any humans who are experts at that level in every field.
Care to elaborate? What is AI psychosis? How am I exhibiting it? I thought hacker news was the last place free of mindless dunking on the internet, I guess I was wrong. If you'd like to engage in a debate on the original topic of this thread I'd be more than happy to, but if you want to dunk, twitter is over at x.com now.
This! It got a much wider pool of templates to copy from, so now if you ask for a 3D web game it gives you a similarly boring game using three.js instead of failing entirely. It still has no imagination and still makes up nonsense all the time. The fundamental problems haven't improved.
This argument style is always humorous. The intention is something like “so humans are as bad as AI” when the original question boils down to something like “why would I replace humans with AI?”.
> The intention is something like “so humans are as bad as AI” when the original question boils down to something like “why would I replace humans with AI?”
If AI really is at human level quality/error rate (I don't think it is for general tasks, but there are some areas where it is), then the answer is typically cost and speed/capacity.
Have outputs from engineers traditionally been measured in cost and speed?
Remember, we aren’t just talking about the product you create. While you would measure deliverables by cost and speed are we ignoring something else? Something that could potentially be more important than either of those metrics?
> Have outputs from engineers traditionally been measured in cost and speed?
Yes. How long it'll take and how much it'll cost are going to be among pretty much any customer's first questions.
They're not the only considerations, and could potentially be outweighed by other concerns even when quality is the same, but I think they are the main drives of AI adoption in industry. If error rate is the same, a $1/hr (amortized) camera and machine vision model capable of checking 300ft of material for defects per minute will likely be preferred to a $10/hr human QA capable of checking 30ft per minute, for instance.
My understanding of your argument is (paraphrasing):
> > People try to excuse AI issues/failure modes by saying humans have them too, but even if they're equally bad then what would be the whole point of replacing a human worker with AI?
To which my response is that speed and cost are also important factors, which can often give AI the edge in considerations when quality/error rate is equal.
If you meant something other than that, you may have to specify.
Speed and cost are nothing without quality and quality is partially a product of accountability (not even considering the technical or logistical issues this is enough on its own.
An AI cannot be held accountable. It does not desire to feed its family.
Your counter argument was outside the context of this articles claims, specifically that programmers and other knowledge workers can be replaced by LLMs.
Equating simple yes/no outcomes generated by vision based machine learning is quite different than “build me a product people will be happy with” being asked of a non-deterministic machine.
Quality was what the hypothetical was assuming had reached parity, no? ("humans are as bad as AI", "If AI really is at human level quality/error rate", etc.)
> accountability
Why could the company not still take accountability? That's already the case for non-ML automated systems, some with high failure rates. As a customer I rarely if ever care about blame being pinned on a specific employee.
> Your counter argument was outside the context of this articles claims, specifically that programmers and other knowledge workers can be replaced by LLMs.
AndrewKemendo's comment and your reply ("any humans", "replace humans") seemed to generalize, but speed and cost being important factors is still true for knowledge work. For some given level of quality, a web developer offering a lower quote with shorter turnaround time will be preferred to one offering a higher quote with longer turnaround time.
For the set of [all tasks humans can do] there is a subset of [things automation can do at equvalent or better error rates].
And that’s the baseline the baseline is not some platonic ideal that is never reached the baseline for a business operator is cost per delivery /error.
The idea that existing human systems are optimized or otherwise “not broken”, is the key fallacy that people keep making.
The entire purpose of automation is to remove a capacity limited human from a continuous workflow because the workflow is more capably achieved with fewer errors than the human
If I have a choice between a deterministic traffic light and a non-deterministic traffic light which one would I use?
And yes, before you say “this isn’t a comparison of non deterministic and deterministic tools, this is a comparison of two non-deterministic tools” think about what my next question might be.
I am unaware of any healthy human who confabulates things as arbitrarily and disastrously as a SOTA reasoning model. It is childish to say stuff like "lawyers always made up court cases" - no they didn't!
That is the entire industry of business consulting.
Boston consulting group Bain and MacKenzie make billions of years completely making shit up. same thing with Ernst and young and any of these organizations that make these “future of (insert market)” reports
Look, I don't spend most of my time online criticizing AI progress. But what does your response even mean? People hallucinating work and solutions isn't commonplace at all, right? What industry do you work in where people hallucinate with frequency?
I can't speak to GP's intention, but I've personally witnessed a guy on my team who was trying to position himself as the go-to technical dude. He was jockeying for a management role. When QA or customer support had questions about our products, he'd always have an answer. I would say that at least 50% of the time, his answer was completely fabricated nonsense. He'd wildly misrepresent projects that his teammates were working on. I also saw several incidents of cargo-cult programming from him. Bizarrely, this never bit him in the ass and now he's a middle manager at a FAANG. This experience leaves me without much hope for the future of software development as a career.
Although it's not the most important thing here, I love the IQ badge on the right side of the screen. At IQ 200, the AI is finally qualified to carry a pager for production.
I'm in the pagerduty lineup, and shockingly, my IQ isn't even a mere 185.
Man the one-shot game that is genuinely good in 2027 is crazy. A good game typically takes around 3-8 years of development by multiple skilled people. Maybe 3 years or so for 1 guy that's super dedicated.
Right now single prompt with Fable can get us a small protype of like 1 game mechanic that's not even remotely production ready. So this guy thinks it'll 100,000X in a year.
It's like any time I ever wanted to make a game. "What if there was a game where you could drive around and shoot people?" Oh it turns out driving a square to throw circles at other squares isn't actually very fun. AI does the same.
I'm having a hard time understanding what the writer is trying to accomplish with this (weird) post, especially the "IQ" part on the side (wtf?). Yes, the goal posts change as the technology changes (this is true of ALL technology), and it turns out that people keep underestimating with each release of this technology what it would actually take to replace a job.
It's a common joke to reply to something with 'yes' that doesn't make sense for the question asked. When used in place of a number, it just means extremely high, as high as it will go, that no matter what number you need it's a 'yes', and so on.
I heard it with budgets first. If your budget is "yes", it means your boss says yes whenever you ask for money. By analogy if your AI's IQ is "yes" the AI can do anything you can think of. It's also possible to have a budget of "no" and this is much more common.
The irony is that while this is all mostly true, the site itself is clearly written with AI. Clearly matters here, because although it can do all of these remarkable things the prose it writes is still completely banal.
As my company experiments with AI-heavy programming, I don’t feel like my job’s going away, but as a “full stack” programmer who’s also good at and comfortable managing a task board, walking “stakeholders” and SMEs through requirements-gathering, et c., I am increasingly wondering why our “teams” still have a couple non-programmers doing that stuff. Dedicated QA, for that matter—all that test-writing and test-data-generating and stuff is so fast now. It’d speed things up if one or two programmers just did the whole thing.
But maybe my particular skill set where all those roles were only really out of reach for me for time-constraint reasons is less common than I think… I dunno, though, between people who’d moved up into managing projects but can drive an LLM pretty competently for programming (ex programmers) and versatile can-talk-to-people seasoned programmers, it’s all the other roles that look increasingly like they’re slowing productivity, rather than increasing it, now.
Someone on a call the other day tentatively brought up that they were noticing it was taking longer to get all the paperwork right for a bug fix or even mid-sized feature than it was to actually write and test the fix, by the time they looped in some variety of person in a jira-wrangler role. It’s clear those jira-wranglers are gonna have to fight to keep their jobs (I don’t really know how they’re gonna do it, I feel apprehensive on their behalf in every meeting now)
In my experience, those roles never existed to begin with in well run teams and companies. Some companies drive product managers into this role, but that was always a bad use of their time. Even with AI tools, I still don't feel like I have the skills to do the job of a good product manager, which requires the kind of vision and business acumen that I am just not as good at as the best PMs I've worked with.
Yeah right after I posted I thought about editing in that a 3-4 all-programmer team with one person (maybe half-time) acting as basically team secretary was already the best unit for software writing, but bigcos seemed to have trouble forming them without adding a bunch of overhead and making their programmers often blocked or idle for some reason.
It’s just that now if I have an LLM that can talk to my business comms and reporting tools (jira was already mostly for managers to generate reports, not to help the people doing the work, that’s why it sucks so much as a tool for getting actual work done) and my first-pass at programming looks a lot like just writing a programmerly-flavored feature or bug ticket, having other people making useless first drafts of tickets for me to rewrite (and probably have to go clarify them with an SME or stakeholder anyway) and someone dedicated to poking around in Jira in general, is gonna start looking funny even to bigcos, I think.
Some people will exaggerate one way, and some the other way. The internet is big and you can say, and substantiate, "people said" for any opinion you choose, and show that it was wrong later.
I think this is a good example of the moving goalposts. The Turing test is not "AI is indistinguishable from human writing 100% of the time", it's whether it is possible at all for a system can be designed where a person can be fooled into thinking they're talking to a human when they're actually talking to a computer in a turn-based text exchange. It is a "there exits..." problem rather than a "for all..." problem.
I think your post is a very good example of goal post moving.
We currently have debates around whether content online is human or ai generated and instead of acknowledging the milesetone we have a post essentially claiming "the turing test was passed long ago, it's worthless".
The one-shot game is definitely the hardest one here by a long shot, no idea why this is considered easier than having a model on pager duty. Aside from needing to generate coherent assets and having a meaningful creative vision, a good game is hard to make because it requires a lot of taste, not being good at programming (see the many indie hits from people who learned to program just to make their game, and were wildly successful with atrocious code).
I hate to say it as I think the author is well intentioned but I think this is every bit as silly as the perspectives it is shooting down
1. this is prone to taking the fringe perspectives and making them "everyone"
2. most of the points are highly highly interpretive (growth could be pressure, laziness, FOMO, self deception, real valid growth.. we don't know yet)
LLMs are definitely the most transformative thing I have seen in 25 years in tech, but I still think like every other hype cycle there is a lot of lying and self delusion, I don't really think any of us, neither myself nor the author really know what we have here.
I like that this goofy fear-based boosterism is on the front page at the same time as an actually well reasoned and well written article about how fear-based boosterism has actually harmed the AI industry instead of making everyone panic-buy like they intended.
I assume that’s entirely due to not being able to downvote submissions on HN.
I have really enjoyed this period of HN, because there has just been a genuine split here on this topic since it became The Thing, which means I get to see a whole ranges of opinions and arguments on this when I come here. I'll be really bummed if either the doomers or the boosters (or the pragmatists) get run off of the site. A monoculture bubble wouldn't be nearly as useful to me.
The AI boosterism is really weird. What is the big deal? My job is not to make AI succeed, it's too solve problems for our clients. Yet I'm getting told I need to use AI to do my job or my funding is going to get cut. We've done several research spikes to figure out where AI can fit into our project and apparently coming up with the answer of, "nowhere near the classified data" that means we must not have AI'd hard enough. My project owner doesn't seem to care if our actual project goals are met, he only seems to care that AI was used.
He's not my boss, so he can't literally tell me what to do. And my actual boss has told me to ignore him. But it's a worrying but of psychosis that I fear could infect the rest of the C-suite of it isn't addressed now.
So I do a lot of client work for F100s and I’ve got some good news and bad.
The bad news is it’s already in the C Suite and they (were) pushing it hard.
The good news is that since Copilot and others have started to charge based on usage a lot of those same leaders have hit the brakes and are now wanting detailed information and usage reporting to figure out where AI actually fits in.
Some have gone even further and slapped a usage limit on individuals or teams and left it at that.
Sanity is around the corner. At least until the next big thing :)
The thing that annoys me most about this obviously AI-generated article is: "The quotes in orange boxes are real and checkable." No they aren't - you AI generated them, just like you generated the rest of the page.
I did notice that the "quotes" weren't linked as I would have expected with that "you can check" language, but after searching for the exact phrases, it does look like they're vague paraphrasing, not actual quotes. Sigh.
Remember: people used to say playing Chess was “AI”. There were dedicated Chess chips, and it was a massive AI undertaking.
We’re treating LLMs like we have every other tool. It’s going to become like spellcheck eventually, not something you think about, and certainly not worth hundreds of billions.
“Moving the goalposts” is a useless criticism. Do I need to pick now a single, definitive goalpost where the massive redirection of resources will have become worth it? Or can I just have a laugh at the dumb state of the present?
I think we're not putting enough emphasis on the importance of human intelligence. Technology is supposed to serve us, we're not supposed to serve technology.
> You can have something smarter than you which you use as a tool.
That sounds like how tech CEO's treat their employees!
Kidding aside, I think the notion that you could control something (legitimately) smarter than you is a pretty risky proposition. (Fortunately one that I think is actually far off)
The ultimate goal of this technology is to replace people -- if it's a tool and corporations still need to employ people to use the tool, then the valuations make no sense. So ultimately, you're going to need to hand the car keys to these agents more and more: here's access to production (we fired the devops people), here's access to all our internal messages and IP, here's all our source code (we fired the engineers).
While I think the discussions around alignment sound more like science fiction than science, it also hasn't been "solved", and the only "control" you have over what the agent does is by gating access. Even if you think the AI itself is aligned, handing complete control of your entire enterprise and all your data to a private unaccountable for-profit enterprise should give you a lot of pause.
The idea that you can create entire jobs where humans just act as reviewers or gate keepers seems very unlikely to me, just knowing human psychology. If the AI is right 99% of the time, and catastrophically wrong %1 of the time, but I've been condition to always hit "accept" because it's usually right, there's very little chance I'm going to catch that 1%. "yes, continue with the database migration". "yes, continue with the database migration." x 1000. (At button click 500, the user stopped reading). "yes, go ahead and delete the database. Shit, wait!!!"
It already has created jobs where people just act as reviewers and gatekeepers though. There are people who vibecoded products and are now running decent-sized businesses.
> "Well, I can still ride a unicycle blindfolded better than a robot so it's not AGI."
Can you though? I've seen some videos recently of some pretty darn scary Chinese robots that I suspect might already be capable of riding a unicycle blindfolded if someone set 'em on that task. ;)
Can you link those videos? What I have seen was impressive but far from "riding a unicycle blindfolded". But with time can be surely done. Also yes, there are likely not many humans who can do it, but most could learn it.
Yeah this idea that the unique value proposition of humans is now our motor capabilities rather than our cognition is unnerving as someone with dyspraxia. Like, oh good, they've figured out how to convert it into a much more limiting disability by commoditising apparently most knowledge work. Great.
If all the AI booster dreams come true with nothing slowing them down, the value proposition for an average human will change to, in this order:
1. At least humans can still do highly specialized physical work
2. At least humans are cheap to run on low-value hard labor tasks
3. At least humans are desperate and highly expendable, in some types of work it's cheaper to 'use one up' and replace them with another than to run and constantly repair robots
4. Humans are useless for all work and therefore have no value
> "4. Humans are useless for all work and therefore have no value"
The part that worries me is when some clown with way too much money and power and time on their hands accidentally really does tip a machine over into AGI and the machines figure this little detail out for themselves... You want "Terminator IRL"? Because that's how you get "Terminator / SkyNet IRL"...
Maybe we can live-stream it all on Pay-Per-View once it starts, to make the ultra-rich idiots who ended us all just a little bit richer before the end, as I'm pretty sure some folks out there actually would pay to watch live-stream feeds of the ensuing apocalypse right up until the machines come to kick their door in too.
It’s not meaningless because AGI = genuinely being able to replace humans in almost any job. Sure, AI is getting better at stuff that humans direct it to do, but the fact that a human is required in the loop is super meaningful.
Maybe I'm wrong, but I think it will always be used as a tool however smart it gets.
In software, it hasn't replaced all the software engineers (even though it's better and faster than us), it's just meant that we now have more leverage.
We can write more code and work on more projects than ever. I don't see why we would suddenly stop using it like that?
"AI babysitter" sounds more like a minimum wage job, not a career path. Outsourcing your primary skill to a machine creates less leverage. I'm convinced this is why CEO's love it -- even though you can't replace software devs, you can hold an axe over their head with the threat.
You should add a parallel timeline of how many times AI CEOs have claimed their next model is too dangerous to release, AGI is months away, or some white collar job will be obsolete within 6 months.
I don’t know anyone in the tech industry who thinks AGI will never happen or that software engineering and white collar jobs can’t be automated. We all read sci fi, you’re not some unique visionary for anticipating AI. The frustration is with how much the claims have outpaced reality and how poorly the investors and executives have treated their workers during this transition.
Yeah reading this timeline wasn't satisfying to me for exactly this reason: It didn't give the counterarguments that were being made at the time of each "goalpost".
My (probably flawed, but still) memory is that the first one of these threads I participated in, at the end of 2022, was saying that none of us would have jobs after two more years of development of these models. Two years from then was almost two years ago now, and we're still "a year or two" out.
On the flip side, the thesis that these will be useful tools that will augment the work of software developers when understood and used for the things they are good at has (IMO) remained undefeated during this entire period.
I’ll go on record and say that AGI will never happen (in the next 50 years). I think that’s also the timeline for white-collar job automation that requires critical thinking.
The article is about goal post moving. Please provide a clear and concrete definition of AGI that we can judge your prediction by.
But the author acknowledged times that someone said something stupid about AI?
He's not claiming there's no flaws or that there aren't CEOs claiming more than is possible right now. He's making the point that these don't negate the parts that are genuinely impressive.
I've encountered waaaay too many "you can't possibly have done that with AI because AI occasionally hallucinates" and "CEOs say things for marketing therefore AI can't really do anything" type of posts.
Understandably, the frustration is also around "what happens then?"
Everyone is wondering about what happens if (when?) it finally comes true
We really have to figure out what comes next for your average person. I think the reason no one wants to talk about this is because the answers aren't great. The average person not going to be living a great life in all likelihood, once we have no access to capital anymore
Yup, that's the weird thing about the booster argument. "We finally achieved it, we're no longer important or the dominant species and Sam Altman and Dario Amodei run what's left of the world from their bunker!" Yay?
Either we achieve a post scarcity society, or we become a permanent underclass if not soylent. Those are the only outcomes I believe are possible.
If we don't have access to capital, we will abolish capital altogether.
This still backfire to the oligarchs
Are you sure the capital-enabled won't abolish us for trying to do so?
They will certainly try, but they don't have the capability yet
It's probably a really bad idea to keep building them autonomous killing machines though
I wouldn't be surprised if multiple nation-states already had them without calling them such because 'Look, someone pushed a button 3 hours ago before it engaged'. Once the private capital gets their hands on them, we immediately transition to cyberpunk no matter what else happens in the world.
Hyper proliferation of drones seems to end in a state of mutually assured destruction, from my armchair.
The average person will not have any life in all likelihood. That is the end goal of capitalism. However, shortly after everyone dies, the remaining survivors - the likes of Musk, Thiel etc - will find themselves roughly equal and needing to establish a new hierarchy.
[dead]
You don’t know me, but I am in the tech industry and believe AGI will never happen.
Albeit that is because I did a BSc in psychology where I developed a deep distrust for intelligence research and concluded that intelligence is not a useful term in philosophy nor science (and especially not in engineering).
I strongly agree. In scientific terms, we already know how to achieve 1G constant acceleration space flight. "All we need to do" from an engineering standpoint to achieve it is develop miniaturized fusion reactors and superconducting electronics. Hell, something approximately as good was achievable since the 1950s with a gargantuan ship that pooped out atomic bombs through a giant shock absorbing pusher plate[0]. Trivial, right? Except it isn't. Both of these projects, which are scientifically feasible, are completely impossible to actually build. They are off the table in engineering terms.
We are so much closer to 1G constant acceleration space flight than we are to AGI. We know, in principle, how to achieve 1G travel. We don't know, in principle how to achieve AGI. Our best guess so far is something along the lines of "emergence" which means "maybe if we do enough matrix math in the right way it'll wake up and become a being with agency and intelligence". Another way to say this is "hopes, prayers, and lots of GPUs".
Let's all get a grip. Without a coherent theory of intelligence, you aren't gonna make one in a lab. That's not how science works, it's not how engineering works. Start at the beginning.
[0] https://en.wikipedia.org/wiki/Project_Orion_%28nuclear_propu...
What’s your definition of AGI?
Strong AGI would be a system capable of terraforming a planet without any external intervention.
I mean, we did it and there have been roughly 117 billion human beings with GI in all of existence.
[dead]
AGI is when an AI model can pass the Turing test. It was achieved with GPT-2.
ELIZA could pass the Turing Test and is over 50 years older than GPT2. You are claiming we have had AGI available since before the first version of Unix came out?
Literally every CEO AI or otherwise claims bullshit about their product. It's their job. It's got nothing to do with people actually using something and not coming to terms with it having any value. If you've never seen a screwdriver it's a waste of time compared to nails.
“Everyone else is doing it” is a childish justification and there’s a line where this becomes fraud and a felony offense. Do you think Elizabeth Holmes should have been allowed to make up claims about their blood testing technology? Is it OK to grossly overestimate Claude’s capabilities when the US military starts using it to determine strike targets?
I'm not justifying anything nor am I arguing against it. I'm saying it's not a thing worth mentioning as it's not related to AI in any way. If one would try to find AI related negatives go for datacentres, chip shortage/price hikes etc... So many better arguments to choose than just saying "CEO's are lying!!1!".
Screwdrivers and nails are deterministic, AI is probabilistic. That is a significant distinction and why your analogy fails.
It's their job to lie and defraud the public? Why don't you believe there shouldn't be consequences for lying and defrauding the public? Do you believe tech CEOs are above society and decency? Does their wealth make them immune from accountability?
Did I ever say any of those things?
IDK why you needed AI to write the little blurbs, it's not really a lot of text.
Like if you feel it's not important enough to write yourself, just don't put anything there?
A lot of people who believed the hype are now actually dependent on LLMs for day-to-day tasks.
They have genuinely been conditioned out of being able to write a sentence without the help of their thinking machines.
It's gotten so bad that I feel that the majority of HN blog posts on the topic of AI are partially or fully AI-generated. Even the ones that are highly critical of the AI industry. People have just spent so much time next to AI, soaking it in, that they're becoming dependent on it - either because they actually can't write for themselves anymore, or because they care about a topic so little that they can only squeeze out a short paragraph and then ask their LLM to fill in the blanks for what were supposed to be their thoughts. For AI boosters it's somehow even worse.
> Call me when it stops making things up.
We haven’t moved past this yet
To be fair I think we'd be able to claim AGI is here if that problem is solved. At this point the models are so smart they're borderline super intelligent if they were cognizant of hallucinations and their own shortcomings. If GPT 5.5 or Opus 4.8 could tell you "I don't know" they'd certainly be "smarter" than any individual human. Some specialists might be better in niche domains, but I don't know of any humans who are experts at that level in every field.
Your condition is called AI psychosis. Good news is: it's curable!
Care to elaborate? What is AI psychosis? How am I exhibiting it? I thought hacker news was the last place free of mindless dunking on the internet, I guess I was wrong. If you'd like to engage in a debate on the original topic of this thread I'd be more than happy to, but if you want to dunk, twitter is over at x.com now.
This! It got a much wider pool of templates to copy from, so now if you ask for a 3D web game it gives you a similarly boring game using three.js instead of failing entirely. It still has no imagination and still makes up nonsense all the time. The fundamental problems haven't improved.
I’m unaware of any humans that don’t have this error method also.
This argument style is always humorous. The intention is something like “so humans are as bad as AI” when the original question boils down to something like “why would I replace humans with AI?”.
We must give this fallacy a name. It’s the facile way out of the argument used by boosters whenever one dares to criticize LLMs.
> The intention is something like “so humans are as bad as AI” when the original question boils down to something like “why would I replace humans with AI?”
If AI really is at human level quality/error rate (I don't think it is for general tasks, but there are some areas where it is), then the answer is typically cost and speed/capacity.
Have outputs from engineers traditionally been measured in cost and speed?
Remember, we aren’t just talking about the product you create. While you would measure deliverables by cost and speed are we ignoring something else? Something that could potentially be more important than either of those metrics?
> Have outputs from engineers traditionally been measured in cost and speed?
Yes. How long it'll take and how much it'll cost are going to be among pretty much any customer's first questions.
They're not the only considerations, and could potentially be outweighed by other concerns even when quality is the same, but I think they are the main drives of AI adoption in industry. If error rate is the same, a $1/hr (amortized) camera and machine vision model capable of checking 300ft of material for defects per minute will likely be preferred to a $10/hr human QA capable of checking 30ft per minute, for instance.
Oh machine learning has been useful for measuring deterministic and non deterministic outputs for a long time.
But that’s not the argument here, is it?
So the question still stands.
My understanding of your argument is (paraphrasing):
> > People try to excuse AI issues/failure modes by saying humans have them too, but even if they're equally bad then what would be the whole point of replacing a human worker with AI?
To which my response is that speed and cost are also important factors, which can often give AI the edge in considerations when quality/error rate is equal.
If you meant something other than that, you may have to specify.
Sure, let me be blunt.
Speed and cost are nothing without quality and quality is partially a product of accountability (not even considering the technical or logistical issues this is enough on its own.
An AI cannot be held accountable. It does not desire to feed its family.
Your counter argument was outside the context of this articles claims, specifically that programmers and other knowledge workers can be replaced by LLMs.
Equating simple yes/no outcomes generated by vision based machine learning is quite different than “build me a product people will be happy with” being asked of a non-deterministic machine.
> Speed and cost are nothing without quality
Quality was what the hypothetical was assuming had reached parity, no? ("humans are as bad as AI", "If AI really is at human level quality/error rate", etc.)
> accountability
Why could the company not still take accountability? That's already the case for non-ML automated systems, some with high failure rates. As a customer I rarely if ever care about blame being pinned on a specific employee.
> Your counter argument was outside the context of this articles claims, specifically that programmers and other knowledge workers can be replaced by LLMs.
AndrewKemendo's comment and your reply ("any humans", "replace humans") seemed to generalize, but speed and cost being important factors is still true for knowledge work. For some given level of quality, a web developer offering a lower quote with shorter turnaround time will be preferred to one offering a higher quote with longer turnaround time.
Correct
For the set of [all tasks humans can do] there is a subset of [things automation can do at equvalent or better error rates].
And that’s the baseline the baseline is not some platonic ideal that is never reached the baseline for a business operator is cost per delivery /error.
The idea that existing human systems are optimized or otherwise “not broken”, is the key fallacy that people keep making.
The entire purpose of automation is to remove a capacity limited human from a continuous workflow because the workflow is more capably achieved with fewer errors than the human
See: traffic lights
That’s a great explanation of automation.
If I have a choice between a deterministic traffic light and a non-deterministic traffic light which one would I use?
And yes, before you say “this isn’t a comparison of non deterministic and deterministic tools, this is a comparison of two non-deterministic tools” think about what my next question might be.
An AI traffic light sounds like an excellent way to kill a lot of people
I am unaware of any healthy human who confabulates things as arbitrarily and disastrously as a SOTA reasoning model. It is childish to say stuff like "lawyers always made up court cases" - no they didn't!
? in a professional setting, my coworkers are just randomly gonna make stuff up
That is the entire industry of business consulting.
Boston consulting group Bain and MacKenzie make billions of years completely making shit up. same thing with Ernst and young and any of these organizations that make these “future of (insert market)” reports
right, so none of my human coworkers ever
Look, I don't spend most of my time online criticizing AI progress. But what does your response even mean? People hallucinating work and solutions isn't commonplace at all, right? What industry do you work in where people hallucinate with frequency?
I can't speak to GP's intention, but I've personally witnessed a guy on my team who was trying to position himself as the go-to technical dude. He was jockeying for a management role. When QA or customer support had questions about our products, he'd always have an answer. I would say that at least 50% of the time, his answer was completely fabricated nonsense. He'd wildly misrepresent projects that his teammates were working on. I also saw several incidents of cargo-cult programming from him. Bizarrely, this never bit him in the ass and now he's a middle manager at a FAANG. This experience leaves me without much hope for the future of software development as a career.
Although it's not the most important thing here, I love the IQ badge on the right side of the screen. At IQ 200, the AI is finally qualified to carry a pager for production.
I'm in the pagerduty lineup, and shockingly, my IQ isn't even a mere 185.
Perhaps that is the true measure that the author found, and humans have actually been winging it since the inception of programming? ;)
But yours is human IQ, theirs is Artificial IQ.
It's like human to dog years ;-)
Man the one-shot game that is genuinely good in 2027 is crazy. A good game typically takes around 3-8 years of development by multiple skilled people. Maybe 3 years or so for 1 guy that's super dedicated.
Right now single prompt with Fable can get us a small protype of like 1 game mechanic that's not even remotely production ready. So this guy thinks it'll 100,000X in a year.
AI boosters are something else.
It's like any time I ever wanted to make a game. "What if there was a game where you could drive around and shoot people?" Oh it turns out driving a square to throw circles at other squares isn't actually very fun. AI does the same.
I'm having a hard time understanding what the writer is trying to accomplish with this (weird) post, especially the "IQ" part on the side (wtf?). Yes, the goal posts change as the technology changes (this is true of ALL technology), and it turns out that people keep underestimating with each release of this technology what it would actually take to replace a job.
What on earth does "IQ: Yes" mean at the end??
It's a common joke to reply to something with 'yes' that doesn't make sense for the question asked. When used in place of a number, it just means extremely high, as high as it will go, that no matter what number you need it's a 'yes', and so on.
I heard it with budgets first. If your budget is "yes", it means your boss says yes whenever you ask for money. By analogy if your AI's IQ is "yes" the AI can do anything you can think of. It's also possible to have a budget of "no" and this is much more common.
The irony is that while this is all mostly true, the site itself is clearly written with AI. Clearly matters here, because although it can do all of these remarkable things the prose it writes is still completely banal.
As my company experiments with AI-heavy programming, I don’t feel like my job’s going away, but as a “full stack” programmer who’s also good at and comfortable managing a task board, walking “stakeholders” and SMEs through requirements-gathering, et c., I am increasingly wondering why our “teams” still have a couple non-programmers doing that stuff. Dedicated QA, for that matter—all that test-writing and test-data-generating and stuff is so fast now. It’d speed things up if one or two programmers just did the whole thing.
But maybe my particular skill set where all those roles were only really out of reach for me for time-constraint reasons is less common than I think… I dunno, though, between people who’d moved up into managing projects but can drive an LLM pretty competently for programming (ex programmers) and versatile can-talk-to-people seasoned programmers, it’s all the other roles that look increasingly like they’re slowing productivity, rather than increasing it, now.
Someone on a call the other day tentatively brought up that they were noticing it was taking longer to get all the paperwork right for a bug fix or even mid-sized feature than it was to actually write and test the fix, by the time they looped in some variety of person in a jira-wrangler role. It’s clear those jira-wranglers are gonna have to fight to keep their jobs (I don’t really know how they’re gonna do it, I feel apprehensive on their behalf in every meeting now)
In my experience, those roles never existed to begin with in well run teams and companies. Some companies drive product managers into this role, but that was always a bad use of their time. Even with AI tools, I still don't feel like I have the skills to do the job of a good product manager, which requires the kind of vision and business acumen that I am just not as good at as the best PMs I've worked with.
Yeah right after I posted I thought about editing in that a 3-4 all-programmer team with one person (maybe half-time) acting as basically team secretary was already the best unit for software writing, but bigcos seemed to have trouble forming them without adding a bunch of overhead and making their programmers often blocked or idle for some reason.
It’s just that now if I have an LLM that can talk to my business comms and reporting tools (jira was already mostly for managers to generate reports, not to help the people doing the work, that’s why it sucks so much as a tool for getting actual work done) and my first-pass at programming looks a lot like just writing a programmerly-flavored feature or bug ticket, having other people making useless first drafts of tickets for me to rewrite (and probably have to go clarify them with an SME or stakeholder anyway) and someone dedicated to poking around in Jira in general, is gonna start looking funny even to bigcos, I think.
Some people will exaggerate one way, and some the other way. The internet is big and you can say, and substantiate, "people said" for any opinion you choose, and show that it was wrong later.
When AI takes my job, I'm dealing illicit contraband to feed my family. There will always be a demand for that.
> Call me when it stops making things up.
We are here.
Yes, it can still produce value.
You forgot the first goalpost: the Turing test.
Even that line's very debatable though. Passing always under every circumstance? No.
Passing specific tests to the point that the internet is now full of "Is that content AI generated or not" debates? Yes
I think this is a good example of the moving goalposts. The Turing test is not "AI is indistinguishable from human writing 100% of the time", it's whether it is possible at all for a system can be designed where a person can be fooled into thinking they're talking to a human when they're actually talking to a computer in a turn-based text exchange. It is a "there exits..." problem rather than a "for all..." problem.
If it's a there exists problem, then it was solved by ELIZA in the 60s.
I think your post is a very good example of goal post moving.
We currently have debates around whether content online is human or ai generated and instead of acknowledging the milesetone we have a post essentially claiming "the turing test was passed long ago, it's worthless".
How am I moving the goal post by pointing out that, by their definition, the goal "was scored" in the 1960s.
That's not a Turing test
It is the original Turing test: https://en.wikipedia.org/wiki/Turing_test#Versions
The one-shot game is definitely the hardest one here by a long shot, no idea why this is considered easier than having a model on pager duty. Aside from needing to generate coherent assets and having a meaningful creative vision, a good game is hard to make because it requires a lot of taste, not being good at programming (see the many indie hits from people who learned to program just to make their game, and were wildly successful with atrocious code).
I hate to say it as I think the author is well intentioned but I think this is every bit as silly as the perspectives it is shooting down
1. this is prone to taking the fringe perspectives and making them "everyone" 2. most of the points are highly highly interpretive (growth could be pressure, laziness, FOMO, self deception, real valid growth.. we don't know yet)
LLMs are definitely the most transformative thing I have seen in 25 years in tech, but I still think like every other hype cycle there is a lot of lying and self delusion, I don't really think any of us, neither myself nor the author really know what we have here.
I like that this goofy fear-based boosterism is on the front page at the same time as an actually well reasoned and well written article about how fear-based boosterism has actually harmed the AI industry instead of making everyone panic-buy like they intended.
I assume that’s entirely due to not being able to downvote submissions on HN.
As for the article, as another user put it:
> Call me when it stops making things up.
We haven’t moved past this yet
I have really enjoyed this period of HN, because there has just been a genuine split here on this topic since it became The Thing, which means I get to see a whole ranges of opinions and arguments on this when I come here. I'll be really bummed if either the doomers or the boosters (or the pragmatists) get run off of the site. A monoculture bubble wouldn't be nearly as useful to me.
The AI boosterism is really weird. What is the big deal? My job is not to make AI succeed, it's too solve problems for our clients. Yet I'm getting told I need to use AI to do my job or my funding is going to get cut. We've done several research spikes to figure out where AI can fit into our project and apparently coming up with the answer of, "nowhere near the classified data" that means we must not have AI'd hard enough. My project owner doesn't seem to care if our actual project goals are met, he only seems to care that AI was used.
He's not my boss, so he can't literally tell me what to do. And my actual boss has told me to ignore him. But it's a worrying but of psychosis that I fear could infect the rest of the C-suite of it isn't addressed now.
So I do a lot of client work for F100s and I’ve got some good news and bad.
The bad news is it’s already in the C Suite and they (were) pushing it hard.
The good news is that since Copilot and others have started to charge based on usage a lot of those same leaders have hit the brakes and are now wanting detailed information and usage reporting to figure out where AI actually fits in.
Some have gone even further and slapped a usage limit on individuals or teams and left it at that.
Sanity is around the corner. At least until the next big thing :)
The thing that annoys me most about this obviously AI-generated article is: "The quotes in orange boxes are real and checkable." No they aren't - you AI generated them, just like you generated the rest of the page.
I did notice that the "quotes" weren't linked as I would have expected with that "you can check" language, but after searching for the exact phrases, it does look like they're vague paraphrasing, not actual quotes. Sigh.
Remember: people used to say playing Chess was “AI”. There were dedicated Chess chips, and it was a massive AI undertaking.
We’re treating LLMs like we have every other tool. It’s going to become like spellcheck eventually, not something you think about, and certainly not worth hundreds of billions.
> The founder
> An AI notices an unmet need, builds the product, finds the customers, and runs the company to a billion-dollar valuation with zero employees.
I'm OK with this. Delamain is awesome.
BEEP BEEP MOTHERFUCKER!
Thank you very much for creating this!
I’m gonna send this to people when they tell me that nothing‘s happening and none of this is real
I love this so much thank you
It does sound like a good way to get people to stop talking to you
I don’t care what it does really. It’s another tool.
It’s the economics that make no sense. That situation has been demonstrably getting progressively worse.
Call me. When the AI stops writing. Every blog post. Like this.
“Moving the goalposts” is a useless criticism. Do I need to pick now a single, definitive goalpost where the massive redirection of resources will have become worth it? Or can I just have a laugh at the dumb state of the present?
This is strawmaning the article
I think this is true of "AGI" too.
AI is already better than us at a bunch of things, and worse at a few. The list of things it's better than us at increases every month.
In 5 years, people will still be saying "Well, I can still ride a unicycle blindfolded better than a robot so it's not AGI."
AGI is such a meaningless term and puts too much importance on human-level intelligence.
I think we're not putting enough emphasis on the importance of human intelligence. Technology is supposed to serve us, we're not supposed to serve technology.
I agree. I don't think they're mutually exclusive. You can have something smarter than you which you use as a tool.
My point was that there's nothing objectively special about our level of intelligence, so it shouldn't be used as a benchmark.
> You can have something smarter than you which you use as a tool.
That sounds like how tech CEO's treat their employees!
Kidding aside, I think the notion that you could control something (legitimately) smarter than you is a pretty risky proposition. (Fortunately one that I think is actually far off)
> I think the notion that you could control something (legitimately) smarter than you is a pretty risky proposition.
But then think about the CEOs and their employees again.
We just need to invent something like money for an "AI" and then we can lead in on a stick ;-)
In how many companies or countries is the smartest person in charge? Raw intelligence is just one of many factors that affect a person’s abilities.
I'm able to control my coding agent even though it writes code better and faster than I can.
The ultimate goal of this technology is to replace people -- if it's a tool and corporations still need to employ people to use the tool, then the valuations make no sense. So ultimately, you're going to need to hand the car keys to these agents more and more: here's access to production (we fired the devops people), here's access to all our internal messages and IP, here's all our source code (we fired the engineers).
While I think the discussions around alignment sound more like science fiction than science, it also hasn't been "solved", and the only "control" you have over what the agent does is by gating access. Even if you think the AI itself is aligned, handing complete control of your entire enterprise and all your data to a private unaccountable for-profit enterprise should give you a lot of pause.
The idea that you can create entire jobs where humans just act as reviewers or gate keepers seems very unlikely to me, just knowing human psychology. If the AI is right 99% of the time, and catastrophically wrong %1 of the time, but I've been condition to always hit "accept" because it's usually right, there's very little chance I'm going to catch that 1%. "yes, continue with the database migration". "yes, continue with the database migration." x 1000. (At button click 500, the user stopped reading). "yes, go ahead and delete the database. Shit, wait!!!"
It already has created jobs where people just act as reviewers and gatekeepers though. There are people who vibecoded products and are now running decent-sized businesses.
> I think the notion that you could control something (legitimately) smarter than you is a pretty risky proposition.
Well, Trump seems to control a lot of people given how afraid they are.
Trump is also called not the smartest person out there.
So...
> "Well, I can still ride a unicycle blindfolded better than a robot so it's not AGI."
Can you though? I've seen some videos recently of some pretty darn scary Chinese robots that I suspect might already be capable of riding a unicycle blindfolded if someone set 'em on that task. ;)
Can you link those videos? What I have seen was impressive but far from "riding a unicycle blindfolded". But with time can be surely done. Also yes, there are likely not many humans who can do it, but most could learn it.
Here's one from a few months ago -> Kungfu robots (o.o) https://youtu.be/UKLvMLtNXpE
You can search and find tons more like that on your favorite YouTube frontend. :)
I can't even ride a bike so I got AGI'd about a decade ago
Yeah this idea that the unique value proposition of humans is now our motor capabilities rather than our cognition is unnerving as someone with dyspraxia. Like, oh good, they've figured out how to convert it into a much more limiting disability by commoditising apparently most knowledge work. Great.
If all the AI booster dreams come true with nothing slowing them down, the value proposition for an average human will change to, in this order:
1. At least humans can still do highly specialized physical work
2. At least humans are cheap to run on low-value hard labor tasks
3. At least humans are desperate and highly expendable, in some types of work it's cheaper to 'use one up' and replace them with another than to run and constantly repair robots
4. Humans are useless for all work and therefore have no value
> "4. Humans are useless for all work and therefore have no value"
The part that worries me is when some clown with way too much money and power and time on their hands accidentally really does tip a machine over into AGI and the machines figure this little detail out for themselves... You want "Terminator IRL"? Because that's how you get "Terminator / SkyNet IRL"...
Maybe we can live-stream it all on Pay-Per-View once it starts, to make the ultra-rich idiots who ended us all just a little bit richer before the end, as I'm pretty sure some folks out there actually would pay to watch live-stream feeds of the ensuing apocalypse right up until the machines come to kick their door in too.
It’s not meaningless because AGI = genuinely being able to replace humans in almost any job. Sure, AI is getting better at stuff that humans direct it to do, but the fact that a human is required in the loop is super meaningful.
Maybe I'm wrong, but I think it will always be used as a tool however smart it gets.
In software, it hasn't replaced all the software engineers (even though it's better and faster than us), it's just meant that we now have more leverage.
We can write more code and work on more projects than ever. I don't see why we would suddenly stop using it like that?
"AI babysitter" sounds more like a minimum wage job, not a career path. Outsourcing your primary skill to a machine creates less leverage. I'm convinced this is why CEO's love it -- even though you can't replace software devs, you can hold an axe over their head with the threat.
I am pretty convinced modern LLMs are AGIs based on the pelicans and ability to write music in MIDI.