(1) trickle-down economics-style policies aren’t the only way to harness the (presumed) increased productivity of LLMs like chatgpt.
(2) I don’t see how you can argue that standards of living have decreased since the start of the industrial revolution and subsequent technological developments.
Certainly, the question of how to distribute the benefits of wide-spread LLM use is very important, since it could be done well or could be done terribly. But there’s no reason to just assume terrible is the only option.
There is no reason to believe that the rich classes will act differently than at any time before in human history. To the contrary, union busting laws have eroded the standing that the worker class had even a few decades ago.
> (2) I don’t see how you can argue that standards of living have decreased since the start of the industrial revolution and subsequent technological developments.
Yes, on a societal whole standards of living have risen, but a) millions of people have suffered regardless (e.g. it has been shown that those graduating during the recent economic crises face significant income losses over their lifetimes) and b) other metrics (ownership of assets like houses, financial savings) have gone dow hill at the same time.
"Trickle down economics" don't work. Industrialization waves wrecked the lives of many many millions of people over the centuries, while the people owning the factories eventually got richer than aristocracy as they kept the price of the sold goods barely cheap enough to be a bit cheaper than manual production and pocketed the immense difference in production cost.
> 2) Chaudry ignores that you need a human in the loop to check an AI's work. You don't know that an AI's answer is always right. Even if it's perfect in 99% of cases, the answer could be jaw-droppingly destructive in that 1% of cases and make errors that even the dumbest human wouldn't make.
Replace "an AI's work" by "a C-level executive". They can run a company fine in 99% of cases and in 1% of cases run it completely aground with extremely dumb short-sighted decisions (like we're seeing with Twitter, or with Boeing, or Credit Suisse), but no one cares about that and the execs still get their golden parachute. It will be just the same with AI - eventual failure will be priced in.
> Did accounting software make the world worse and destroy jobs even though it made 1 accountant ridiculously more productive than they were in the past?
The switch towards computerization wrecked the Western middle classes. It used to be the case that white collar jobs could feed a whole family comfortably on one income, and automation made it so bad that the majority of our populations has to struggle to make ends meet now, and completely forget about having children because we can't afford large enough homes.