It's not "AI bubble that will crash" but definitely some kind of deflation or correction that will damage a lot of investments. As always, common people will be most affected.
It's not "rogue ASI will kill us all" but advancements in cyberattacks, skill erosion, job loss that will damage us all. Again, common people will be most affected.
It's been 5-6 years with these things, how does anyone still believe that this will usher a new era of prosperity? Maybe for Sam and Dario.
If AI keeps advancing it will eclipse humans in all domains. This has been predicted for over a hundred years and it is a reasonable extrapolation from current trends.
Existing among machines more capable than humans in all domains is not a stable outcome. Either AI is significantly curtailed and restricted, or humans are consumed by the machines or disassembled for spare carbon.
Until models have taste, we still need Software Engineers to judge their output in the context of a broader goal. Not saying this will never happen, but we have yet to see evidence of models getting good at this. If anything, the lack of improvement on that axis has been quite underwhelming; the latest example of that being OpenAI offering a Lean proof of the solutions to the Navier-Stokes equations, but that proof not matching the natural language of that proof [1].
OK, but in my experience, ~90% of the software dev economy is CRUD and boilerplate. A very tiny percentage of the dev economy was ever doing anything novel, if you really think about it.
So, this is an interesting question. If "taste" still gives advantage for now, what does that do to the rest of the boilerplate dev economy, which I am still sure was >90% of the entire thing?
What does "taste" even mean? The only taste that matters is PMM taste. Coding is solved, unless you are doing something truly novel.
So, if we can discount 90% of the boilerplate devs, and the 10% can just use tokens.... I mean is anyone denying that we are doing this right now, and killing the entire jr dev ladder???
Am I still really having this conversation right now in 2026? Isn't all of this extremely obvious by now?
Technical ability != building something people want, will use on a daily basis, and provides economic value or enjoyment for humans. Models have been trained to be superhuman at the former, not the latter. Taste is what allows humans to do the latter.
Therefore, until taste can be programmed as a reward function in RL, or we have demonstrated that scaling RL improves model generalisation to such an extent that we don't need to create that reward function,
> Coding is solved, unless you are doing something truly novel.
Agreed. And even then, AI are sometimes better at humans at finding solutions to problems humans can't (e.g. Navier Stokes equations). But that still doesn't mean all tasks of the job are done (though obviously the definition of the job have and will continue changing as the tech improves).
> Technical ability != building something people want, will use on a daily basis, and provides economic value or enjoyment for humans. Models have been trained to be superhuman at the former, not the latter. Taste is what allows humans to do the latter.
I think finally learned of the role title and the audiobook to teach me about this. It will also be replaced, but one of the last "tech" roles to be automated. It's where the taste lives, right?
Role: Product Marketing Manager - with a tech background and agentic dev skills (although the last part is most fleeting)
Book: Obviously Awesome
If anyone else would care to educate me about this here, please and thank you!
Copium isn't a rational response - it's an emotional one, and it's putting yourself in a worse position than if you view things more clearly and can plan accordingly.
IMHO, the US is already cooked. Technological superiority is one of many ways to exert influence, and foreign initiatives are closing in, fast. There's nothing the US can do do stop it, unless it can find a way to stall China's advances in the area, and this is incredibly unlikely.
How are you defining “AI bubble”? Via the capability trajectory of AI or as an economic analysis. How is “AI bubble” an extreme? Was dotcom bubble an extreme? In retrospect it feels like a median position. Bubbles are super normal not extreme. “AI is useless and will go away” would be the extreme position. Your third possibility could happen in the “AI bubble” world.
I think “AI bubble” gets framed as a technological question about AI capabilities and not a financial one. The bubble may deflate slowly, but if you eat all software or eat all other Field’s something is popping. Anthropic OpenAI and SpaceX are all valued partially as if they’ll be the only ones to succeed.
Perhaps look at the past dozens of discussions on this general idea.
One would hope that there were many thousands of this exact discussion. If you could point me to your favorites, I would appreciate it.
I might be a bit pessimistic but not extreme.
It's not "AI bubble that will crash" but definitely some kind of deflation or correction that will damage a lot of investments. As always, common people will be most affected.
It's not "rogue ASI will kill us all" but advancements in cyberattacks, skill erosion, job loss that will damage us all. Again, common people will be most affected.
It's been 5-6 years with these things, how does anyone still believe that this will usher a new era of prosperity? Maybe for Sam and Dario.
If AI keeps advancing it will eclipse humans in all domains. This has been predicted for over a hundred years and it is a reasonable extrapolation from current trends.
Existing among machines more capable than humans in all domains is not a stable outcome. Either AI is significantly curtailed and restricted, or humans are consumed by the machines or disassembled for spare carbon.
I think that you might still be going with the exciting extremes, p(doom) style.
What about p(economic-collapse)? i.e. p(boring)
Where is the discussion by economists on the latter, as that appears to be exactly where we are headed, nearly for sure?
The Economic Nobel exists after all. Shouldn't there be a very public fight over the future on this topic?
> will eat all software and software-related meetings.
Software engineering is a job. Writing code is a task. The task isn't the job.
copium
Until models have taste, we still need Software Engineers to judge their output in the context of a broader goal. Not saying this will never happen, but we have yet to see evidence of models getting good at this. If anything, the lack of improvement on that axis has been quite underwhelming; the latest example of that being OpenAI offering a Lean proof of the solutions to the Navier-Stokes equations, but that proof not matching the natural language of that proof [1].
[1] https://arxiv.org/html/2610.08144v1
OK, but in my experience, ~90% of the software dev economy is CRUD and boilerplate. A very tiny percentage of the dev economy was ever doing anything novel, if you really think about it.
So, this is an interesting question. If "taste" still gives advantage for now, what does that do to the rest of the boilerplate dev economy, which I am still sure was >90% of the entire thing?
What does "taste" even mean? The only taste that matters is PMM taste. Coding is solved, unless you are doing something truly novel.
So, if we can discount 90% of the boilerplate devs, and the 10% can just use tokens.... I mean is anyone denying that we are doing this right now, and killing the entire jr dev ladder???
Am I still really having this conversation right now in 2026? Isn't all of this extremely obvious by now?
Technical ability != building something people want, will use on a daily basis, and provides economic value or enjoyment for humans. Models have been trained to be superhuman at the former, not the latter. Taste is what allows humans to do the latter.
Therefore, until taste can be programmed as a reward function in RL, or we have demonstrated that scaling RL improves model generalisation to such an extent that we don't need to create that reward function,
> Coding is solved, unless you are doing something truly novel.
Agreed. And even then, AI are sometimes better at humans at finding solutions to problems humans can't (e.g. Navier Stokes equations). But that still doesn't mean all tasks of the job are done (though obviously the definition of the job have and will continue changing as the tech improves).
I believe that we generally agree.
> Technical ability != building something people want, will use on a daily basis, and provides economic value or enjoyment for humans. Models have been trained to be superhuman at the former, not the latter. Taste is what allows humans to do the latter.
I think finally learned of the role title and the audiobook to teach me about this. It will also be replaced, but one of the last "tech" roles to be automated. It's where the taste lives, right?
Role: Product Marketing Manager - with a tech background and agentic dev skills (although the last part is most fleeting)
Book: Obviously Awesome
If anyone else would care to educate me about this here, please and thank you!
Copium seems like the rational response to the loss of millions of job, stability, and the meaning of one’s life.
Billions of people are going to become impoverished and your reaction is to insult someone for grieving?
What do you think is the purpose of life? Is this how you’ll react as one by one the souls of the ones you love most go dark?
Copium isn't a rational response - it's an emotional one, and it's putting yourself in a worse position than if you view things more clearly and can plan accordingly.
Against superhuman AI, people are planning accordingly for their irrelevance.
What’s the point of shouting and spreading pessimism?
Do you honestly think writing code is the purpose of life? Do you think meaningless drudgery is the meaning of life?
Yes. Please enlighten me otherwise.
my 2c:
IMHO, the US is already cooked. Technological superiority is one of many ways to exert influence, and foreign initiatives are closing in, fast. There's nothing the US can do do stop it, unless it can find a way to stall China's advances in the area, and this is incredibly unlikely.
How are you defining “AI bubble”? Via the capability trajectory of AI or as an economic analysis. How is “AI bubble” an extreme? Was dotcom bubble an extreme? In retrospect it feels like a median position. Bubbles are super normal not extreme. “AI is useless and will go away” would be the extreme position. Your third possibility could happen in the “AI bubble” world.
I think “AI bubble” gets framed as a technological question about AI capabilities and not a financial one. The bubble may deflate slowly, but if you eat all software or eat all other Field’s something is popping. Anthropic OpenAI and SpaceX are all valued partially as if they’ll be the only ones to succeed.
[flagged]
[flagged]