AI is a bubble and it will burst. Most of the companies will fail. Most of the datacenters will be shuttered or sold for parts. So what will be left behind? We will have a bunch of coders who are really good at applied statistics. We will have a lot of cheap GPUs, which will be good news for, say, effects artists and climate scientists, who will be able to buy that critical hardware at pennies on the dollar. And we will have the open-source models that run on commodity hardware, AI tools that can do a lot of useful stuff, like transcribing audio and video; describing images; summarizing documents; and automating a lot of labor-intensive graphic editing – such as removing backgrounds or airbrushing passersby out of photos. These will run on our laptops and phones, and open-source hackers will find ways to push them to do things their makers never dreamed of.

• This post-bubble future is kind of what I'm looking most forward to, assuming the crash doesn't put me out on the street.

• This, when setting up a web server was still a huge ceremony.
• Now, almost 30 years later, I can run a web server on a microcontroller smaller than my fingernail that cost me less than a buck.
• In fact, I embedded one into a pumpkin for fun, about 8 years ago.
• What if LLMs don't even improve radically, but gradually shed all the ceremony for running them on cheap personal devices?

• This, after reading "Don't fall into the anti-AI hype" from antirez
• I keep thinking about this and I think it's true.
• Specifically, two things can be true at the same time:
  • "AI" as it currently exists is a bubble and most of the high-flying companies pushing slop are going to die messily.
  • LLMs that generate code are going to be in the programming toolkit for a great many folks from here on, indefinitely, like calculators and compilers.
• Remember the dot-com crash back in the 2000s? Well, neither the internet nor the web went away. We built Web 2.0 atop the dark fiber - and that could happen again for "AI". Maybe.
• Look, I know technological inevitability is a myth - this stuff isn't self-executing, it still takes people to build it and carry it forward by choice.
  • But, like the internal combustion engine and jet aeroplanes, there are a lot of folks who find LLMs convenient & productively useful - even if there are measurable harms and perils in their use.
  • See also: smart phones, social media, same-day delivery, plastics, antibiotics, and eating meat.
  • While none of these are inevitable, I think more folks than not see more benefit than not. And, thus, it's unlikely we'll swear off them cold turkey anytime soon.
• Like Dave says, "We get so mired in the question of should we do this -- well we're doing it, time to start looking at the next set of questions."
  • This is why, for example, I don't think it's entirely off-into-the-weeds for a company like Mozilla to aim at bending the trajectory of this stuff, rather than trying to halt or ignore it.
• For what it's worth, I'm not trying to sell anyone on this stuff. I think, where it's genuinely useful, it sells itself.
  • To the extent that I talk about it is mainly me learning out loud.
  • And, honestly, being serially enthusiastic about a shiny object as is my wont.
• But, still, I think "AI" is in a space of way higher value than some folks place it.
  • And yet, orders of magnitude lower than so many CEOs are hyping it.
  • I hope it someday settles down as normal technology, and I think many of us are hoping for that.
  • But, again, it's not going to just evaporate. Not even if the bubble pops.
• Anyway. These are thoughts I've had and I felt like brain-dumping them today, like you do on a blog.