I don't think it'll contract. The people dumping their money into AI think we are at end of days, new order for humanity type point, and they're willing to risk a large part of their fortune to ensure that they remain part of the empowered elite in this new era. It's an all hands on deck thing and only hard diminishing returns that make the AI takeoff story look implausible are going to cause a retrenchment.
You don't think it will contract just because rich people have bet so much on it that they'll be forced to throw good money after bad? That's the only reason?
I don't think it'll contract because I don't think we'll get a signal that takeoff for sure isn't going to happen, it'll just happen much slower than the hypers are trying to sell, so investors will continue to invest because of sunk costs and the big downside risk of being left behind. I'm sure we'll see a major infrastructure deployment slowdown as foundation model improvements slow, but there are a lot of vectors for optimization of these systems outside the foundation model so it'll be more of a paradigm shift in focus.
It's probably exacerbated by the fact that everyone invest money now, I get daily ads from all my banking apps telling me to buy stocks and cryptos. People know they'll never get anywhere by working or saving, so they're more willing to gamble, high risk high reward, but they have nothing to lose
Yeah it wouldn't be a bubble if it didn't have that mentality. Every bubble has had that thought and it's the same now. Kind of hard to notice it though when you are in the eye of the storm.
There were people telling me during the NFT craze that I just don't get it and I am dumb. Not that I am comparing AI to it directly because AI has actual business value but it is funny to think back. I felt I was going mad when everyone tried to gaslight me
The final AI push that doesn't lead to a winter will look like a bubble until it hits. We're realistically ~3 years away from fully autonomous software engineering (let's say 99.9% for a concrete target) if we can shift some research and engineering resources towards control, systems and processes. The economic value of that is hard to overstate.
This isn't a comment on timelines, but a Waymo going wild is going to run over and kill people, so it makes sense to be overly conservative with moving forwards. Meanwhile, if someone hacks into a vibecoded website and deletes everything and steals my user data, no one's getting run over by a car.
Sure. The point I was trying to make is that we can see a technology that is amazing, and seemingly does what we want, and yet has so many edge cases that make it unviable commercially.
Have you tried Comma.ai? While Waymo and Tesla are trying to make fully self-driving autonomous taxi-grade AI and are taking forever to deliver, geohot casually makes the box that we all really want. A device that hooks into your car, and handles left right gas and brakes on the freeway. I can handle the driving between my house and the freeway, and the freeway offramp to my destination, it's the stop and go traffic on the freeway or just sitting there for hours that sucks ass.
We can see a technology and its shortcomings and people will still pay for it. Early cars were trash, but now look where we are.
