It's big, but it's controlled by humans and how society uses it, and to whom the benefits accrue are ultimately political choices. I think this should end with

It's big, but it's controlled by humans and how society uses it, and to whom the benefits accrue are ultimately political choices. I think this should end with "write a letter to your congressperson and senators because the companies building this and current large employers already have lobbyists." I also think we can't be too breathlessly optimistic about where this can go. Producing feasible text was apparently easy. Producing code that meets a prose specification took longer, but we're basically there. But you can know something about the quality of that output by reading it, or reading and running it. But some stuff will be limited by slow real-world processes. > The upside, if we get it right, is staggering. AI could compress a century of medical research into a decade. Cancer, Alzheimer's, infectious disease, aging itself... You can't really do medical research on aging dramatically faster than people age. Maybe there's a metabolic process which, if you started intervening before people hit 60, would help more of them live to 100+ -- but more and smarter AI may still require 40+ years to run the experiment.