This is a thought bubble I wrote in 2023. Just two years ago. I still don’t think we’re anywhere close to AGI, but also, I wouldn’t be surprised it if just suddenly appeared and took over all our computers.
We’re at an inflection point. Probably past the initial turn and into the upward swing. A combination of available hardware and the refinement of decades of machine learning research and iteration on software implementations, we have this.
The birth of AI.
No, it’s not *that* kind of AI yet. General intelligence. It’s a combination of natural language processing, “large language models” or LLMs and deep learning systems for image processing. The Deep Mind stuff that came out of Google and became “stable diffusion” and the host of derivatives.
These are the early days.
The outputs from these machines are still crude, occasionally surprising. You don’t have to dig too deep to find the edges. But also, you get to keep iterating.
Iterating is a recurring thing with these machines.
They’re iterating very quickly on new models, and soon, the machines themselves may be doing the iterating. They will soon be building the new machines that replace the current generations.
and that’s how you get to *that* AI. The big one. Or more likely, the several big ones and a bunch of little ones.
We’re at the point when this is going to happen sooner than we thought it would previously. The 25-50 years prediction is in the past.
It’s coming soon.