It's peak delusion to think that AI is going away.
AI will not crash, but, a slowdown in AI could very well cause a general recession, precisely because it's he only thing holding the economy up.
That's a totally different thing though.
Even if it's not going away, it'll be a breath of fresh air when it's not shoved into every possible place as a marketing tactic, even when it's not necessary or beneficial. AI does have it's place, I just hope we realize soon how every place is not that place.
It will follow the same pattern that technologies like phones, computers themselves, the internet, and (I guess) cloud storage now touch on nearly everything.
Example: even if you're an old-fashioned boutique leather artisan who makes handmade bespoke horse saddles in 2026 like it was 1783... your business also probably uses all that other stuff in some capacity. You probably have some computers for design and payroll and answering emails, a website that lives somewhere in the cloud, etc.
On a closely related note, if you're tired of hearing "AI, AI, AI" everywhere, there's reason to feel encouraged. Ten years ago, nobody would shut the fuck up about "the cloud." It didn't go away, it just matured and now it's how most people do many things.
They're useful certainly and won't be going away after any financial bubble bursts, whether they are so useful they need to be rammed into every facet of our lives remains to be seen.
One big difference between the internet and LLMS - people enthusiastically adopted the internet and found all sorts of uses for it. LLMs companies are mandating usage by employees and trying to force end users to use it, which seems really strange, why would they have to do that if it's so useful?
Companies are 'mandating' use of AI because they see the potential in it.
In 1997, people were not sophisticated enough to grasp how broad the web would be - if they did - managers would have been pushing everyone to adopt the browser and email.
That said, in the 1990's they were actually pushing everyone to adopt 'Windows' etc..
We have a better grasp of how tech affects the economy, and so now there's a lot of hype. Probably too much.
AI is just in 'Phase 1'. The current Chat Bots are basically like the 'Commodore 64' - a 64K computer.
The hype is annoying - I use AI all day and find it useful and drives me crazy but at the same time 'it's real at the baseline'.
It will not 'explode' I think any time soon, it's going to take years, but in 5 years we will look back and see a huge amount of change.
In 5 years, up to 1/2 of AWS services will be AI based for example. All sorts of new features.
An example: RAG is being embedded into most DBs now, that's a big deal and taht's just 'embedding' not even really AI. In a few years, regular 'javascript jockeys' will be used to that kind of thing.
I'm by no means AI-pilled, and I'm skeptical of all these shenanigans ... but 'it's real'. This is happening on about the scale of the Web Browser.
AI has already had a massive impact (machine learning to recognise pictures, spot patterns in data etc) and will continue to do so.
We're talking about LLMs here and the huge amount of money that has been pumped into them by companies like Nvidia in circular deals with their customers. Personally I think the jury is still out on whether LLMs will develop much further than they have. Comparing Opus 5.5 to Fable 5.2 this week I've seen little difference. We'll see in time though.
AI is going to make it's way into everything, it's inevitable.
I lived through internet, .com, mobile, SaaS - I'm telling you from being 'inside' all of those, that LLMs alone are enough to be that impactful.
The are where it will be most visceral is robotics.
Once the robots start cleaning your bathroom, it will seem clear how it's bigger than browsers.
Just because LLMs are useful doesn't mean it will have a trillion dollar evaluation.
An economy built on a lie sooner or later crashes into the reality. The sooner the less damage it will cause.
I don't think demand for Frontier AI will go away either - even as in the near future 90% of AI will be 'commodity AI' - the Frontier foundries will still have competitive differentiation to offer.
Just like SaaS - technically almost all SaaS should be 'commodity' ... and yet it's not.
I think we can all agree that we are 'sick of the hype' but that doesn't mean the engine will diminish.