Gates joke here, maybe "Bryson at the Gates of Dawning AI" or "You really need to foot the bill, Bill"

A very quick response to https://www.gatesnotes.com/home/home-page-topic/reader/a-turbulent-ai-era-and-critical-choices-to-make maybe I'll get around to polishing this, but I was prompted by a journalist; I should be working on finalising my governance of AI courses syllabi today (finalise – ha ha ha)

  1. "genAI is the only real AI" trope – still trying to ring fence from regulation all the AI actually paying the bills, and offer up to governments the AI running bills up.
  2. mentions inequality. If he cares, he'll mention redistribution at some point
  3. we'll fix the AI errors – plays well with trying to prop up genAI valuations, cf my bluesky thread on why that happens
  4. good points about being able to use existing household devices, bad point about "AI adapting to us" as if bad AI is users' fault
  5. "this time is really different" just saw a talk by Svetlana Rujin, European Central Bank, about whether genAI is a GPT: General Purpose Technology like electricity, the internet, and silicon wafers 
    1. do you see what the people who named the GPT in chatGPT did there? cf. getting money from investors point 3 above.
    2. the question the ECB researcher is working on is whether genAI is really discontinuous with the internet (or the internet with telegraphs), and there's no evidence so far it is.
  6. "credible plan for slowing down AI" how about regulating that corporations get charged a fortune if they don't ensure their products are governable, maintainable, and safely usable? 
  7. He mentions the great depression without mentioning that it was caused by financial policy, not technological change (other than to the extent that concentrated power and caused the financial policy)
  8. He mentions AI working for AI which is incoherent, breaks justicre and government, and therefore increases problems not solutions
  9. He's absolutely right about crime, cybersecurity, and actually also physical security e.g. drone war. But again, with good governance and adequate taxes we could keep ahead of that arms race too, also using AI.
  10. He's right about under-regulated governments deploying autonomous weapons against people without having to get other people onboard. 
  11. He's right about kids and AI. I wrote about this in 2010 I got asked to update it a couple years ago for The Love Makers (which I strongly recommend), then Wired republished a better-edited version again.
  12. About his plan:
    1. First part is good, but needs financing and anyway most countries have already started – pretending they haven't reduces the efficacy of that work.
    2. second part
      1. The real "human reserved" jobs are responsibility for every organisation and endeavour. Here's my recent paper about that in the context of government but it extends to anything. 
      2. those person-on-person elderly care workers he talks about instead are literally breaking their backs in an economy that does not scale at all. The billion-scalers industries have to diffuse enough value to keep people not only gainfully employed but adequately technologically enhanced so all jobs have dignity. Not just create a HG Wells/Metropolis style underclass with permanent sh*t jobs.
    3. Third part, he's right to be thinking about redistribution! But wrong about how to do it. He's trying to push the taxes onto robots which are uncountable. Tax the people making the money, not the people using the services or the machines themselves.
I have a better plan that solves more problems more reliably (and I need to get back to writing my book about it – right after I get back to updating my AI governance courses' syllabi.) If anyone knows Bill Gates (and thinks he'd listen) I've got a more formal 2-page version of that plan in bullets I wrote for a different, more liberal tech billionaire (who hasn't picked it up yet either, AFAIK.)

Thanks Chris for making me read this now. 

Comments

Post a Comment