> 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!
It Still Can't Do My Job: Four Years of Moving Goalposts (2022–2026)
121–130 of 143 posts
Re: It Still Can't Do My Job: Four Years of Moving Goalposts (2022–2026)
#122“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?
Re: It Still Can't Do My Job: Four Years of Moving Goalposts (2022–2026)
#123You 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. Th…
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 a…
It’s true that AI hasn’t replaced us all in the last few years, but it’s completely changed software engineering, and honestly a LOT of that change has been in the last nine months or so. And it’s supposedly destroyed the market for juniors.
In the grand sweep of history, does it really matter if AI takes two years to replace us or ten years to replace us?
Re: It Still Can't Do My Job: Four Years of Moving Goalposts (2022–2026)
#124Earlier quoted context omitted.
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.
Much of HN has devolved into Yahoo Answers levels of inane buffoonery, sadly. Especially when it comes to screeching about how AI is so terrible, blah blah. So tiresome.
Re: It Still Can't Do My Job: Four Years of Moving Goalposts (2022–2026)
#125You forgot the first goalpost: the Turing test.
ELIZA passed the casual Turing test
Re: It Still Can't Do My Job: Four Years of Moving Goalposts (2022–2026)
#126Earlier quoted context omitted.
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.
Google's definition is sufficient[0]. I'd also go so far as to say that LLMs aren't even AI, but that can be a bit pedantic. I align with Richard Sutton's reasoning, in that LLMs (and GPT models more generally) aren't "intelligent" in the same way that RL models are[1].
[0] - https://cloud.google.com/discover/what-is-artificial-general...
Re: It Still Can't Do My Job: Four Years of Moving Goalposts (2022–2026)
#127Earlier quoted context omitted.
The article is about goal post moving. Please provide a clear and concrete definition of AGI that we can judge your prediction by.
Google's definition is sufficient[0]. I'd also go so far as to say that LLMs aren't even AI, but that can be a bit pedantic. I align with Richard Sutton's reasoning, in that LLMs (and GPT models more generally) aren't "intelligent" in the same way that RL models are[1]. [0] - https://cloud.google.com/discover/what-is-artificial-general... [1] - https://www.youtube.com/watch?v=21EYKqUsPfg
how is GPT more general than LLM? also it's not really "LLM vs RL", they're kinda orthogonal
Re: It Still Can't Do My Job: Four Years of Moving Goalposts (2022–2026)
#128Earlier quoted context omitted.
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.
Much of HN has devolved into Yahoo Answers levels of inane buffoonery, sadly. Especially when it comes to screeching about how AI is so terrible, blah blah. So tiresome.
I feel the same about the endless hype, so I guess it's in the eye of the beholder
Re: It Still Can't Do My Job: Four Years of Moving Goalposts (2022–2026)
#129Earlier quoted context omitted.
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.
> think about what my next question might be
or you could just say it yourself instead
Re: It Still Can't Do My Job: Four Years of Moving Goalposts (2022–2026)
#130Man 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.
> 100000x in a year
why not? fable 5 (2026) is an inf% improvement over having nothing (2022)
in any case, stop pretending that it can be represented by a single scalar value, or at least use log scale units