[woeirua]

It's not an affordability crisis, it's a financial crisis. The models get cheaper super fast. By this time next year Fable 5 will cost less than Sonnet does today. That's not the problem. The problem is that many companies are going to realize that they don't get any ROI from AI. Generating code faster != more profit. Most of the Fortune 500 will likely realize this and then the token budgets will come crashing down. Most of their ideas are _bad_ ideas. Implementing bad ideas faster, won't lead to more profit.

Sure, you can use AI to potentially replace software engineers, but the F500 are also terrified of not having accountability or making mistakes. They won't be firing any engineers. In that scenario, there's just no room for AI usage. If you have to be responsible for all the code, then... AI has to either manage it completely autonomously (which even Fable can't) or... humans have to be in the loop which means they still have to understand the code. The best way to understand the code is to write the code yourself. So there's no productivity gain to be had.

I'm pro-AI, but I think we're due for a big crash next year.

[Supermancho]

> The models get cheaper super fast. By this time next year Fable 5 will cost less than Sonnet does today.

I'm not sure that's something to rely on. I would be Fable 5 will be phased out and the bleeding edge will be priced up.

[jerf]

I don't know about everyone else, but if you told me that for the next couple of years I, and everyone else, would have only Opus 4.5, I wouldn't exactly cry about it. Especially if in the meantime it got cheaper and maybe a bit faster.

My desire for the latest and greatest continues on, but my need for it in order to get any value out of it at all is much, much smaller. The in-practice delta between all the versions since 4.5 have been much more subtle than, say, the models available a year before Opus 4.5 and Opus 4.5.

The bleeding edge is going to have to earn its price delta. They can't count on me wanting to upgrade just to get something halfway decent anymore.

[N_Lens]

The interesting point is when it becomes smart enough to completely replace 99% of knowledge workers.

[FromTheFirstIn]

Buddy it’s not even going to replace 1%

[stingraycharles]

I’m not entirely sure, it’s my understanding that later this year a lot of compute will go online with the latest hardware that will be significantly cheaper for inference.

The problem is rather, I think, that people always want to use the latest and greatest models. And that training is super expensive.

Potentially we’ll just see less new model releases.

[KolibriFly]

I feel like this is way too binary. I don't have to write every line of code myself to understand the system. I don't write my own compiler HTTP stack or database either

It's more about the level of abstraction. If AI handles 80% of the grunt work and I spend my time on architecture and reviews that's still a win

[asdff]

This works for you because you were trained in The Old Way.

Consider the people younger than you. Who are literally shutting their brain off so AI can cheat on their essays and exams. They aren't going to be good architects or code reviewers.

[vjvjvjvjghv]

That’s a problem. I work quite a bit with AI now by if I didn’t have a ton of experience writing code myself things would go off the rails quickly. I often have to steer the AI in a different direction from the initial plan it comes up with.

[HDBaseT]

This is some "old man yells at cloud" nonsense.

There will always be proficient programmers who can read and write good as good as yourself. I have 0 reason to believe programmers coming out of University today are any less competent than when you came out of University.

[asdff]

Stanford is proctoring now because current day students could no longer be trusted to not cheat. UC professors are asking for SAT to be reinstated because students aren't prepared well and classes had to get dumbed down to be more remedial on these shortcomings they are bringing.

They are absolutely getting worse. Ask a teacher, at any level.

[snackerblues]

How about the reason he just explained? Do you disagree that relying on AI to make changes without knowing what it's doing makes you a worse programmer? How would it not?

If you don't know what a good or a bad approach to any problem is because you've never actually had to solve it, you can't even know what being a proficient programmer means, so how could you hope to be one?

[woeirua]

This is all fine and dandy until a single line of code somewhere in the stack causes a production outage and management wants to know why it happened. Management won't accept you shrugging your shoulders and saying "the AI did it" for very long.

[hhjinks]

If my job devolves to reviewing AI code, I'm becoming a plumber.

[whimsicalism]

Do they get ROI from additional workers? Honestly it is difficult for me to imagine white collar companies that have positive marginal profit of labor but not positive marginal profit of tokens, especially in the future.

[zdragnar]

Considering the massive layoffs and cold market post COVID, the answer is "no" for a lot of companies.

Theoretically, there's a lot of room for marginal work where developer time isn't worth the cost for the output but tokens are cheap enough to make it worthwhile. Very little of that work ends up being customer facing though, so it isn't actually a growth opportunity for the company.

[woeirua]

Well... Elon showed with Twitter that there's probably a large number of companies out there that have had negative ROI on employees too for a long time. Most companies would be hard pressed to show concrete value from each marginal employee.

[sevenzero]

>The models get cheaper super fast.

Weird, why didn't my subscriptions decrease in price then? Oh wait..

[ranyume]

What do you mean? My deepseek subscription got really cheap.

[simianwords]

Its very interesting how you are contradicting the whole article's axioms and then arriving at the same conclusion that we are in for a crash!

Rational takeaway is to step back and analyse what's really happening here.

- Are we really in for a crash?

- What does it say about the culture and people's mental models that we have two radically opposing viewpoints on AI costs and people still arrive at same conclusion?

[fluoridation]

The fact is that a lot of people think AI is overvalued to the point there's no way it can deliver on the hype, so (if they're right) there necessarily must come a crash that brings sentiment down to the same level with objective reality. There's a lot of hype surrounding AI, so there's many different ways to defend the idea that it's overhyped, the same way that if you ask ten different people what is most wrong with the country, you'll get ten different answers.

>- Are we really in for a crash?

The question you should really be asking is, is AI really overvalued, or is it so useful it justifies all the hype that surrounds it? If the former, then yes, a crash is inevitable, because we don't live in the land of make-believe. If a crash never happens then AI was not overvalued, it was valued appropriately.

[whimsicalism]

It's just schadenfreude/ressentiment. Our societal values are very christian even as we secularize and you can see this reflected everywhere you turn. Same thing with people feeling there ought to be a 'catch' to GLP1s, etc. etc.

[simianwords]

I don't get you because I'm too stupid to parse your thoughts

[whimsicalism]

thanks for the productive reply

[simianwords]

I thought you would explain what you meant but I asked Claude and I kinda understood. Interesting analysis.

[woeirua]

You can think that AI is a transformative technology and still come to the conclusion that the AI industry's economics don't pan out. I do think that we will eventually develop AGI. But I think the current scale up economics won't get us there before the music stops.