AI development cannot be left to market forces alone, U.N. experts warn

AI development cannot be left to market forces alone, U.N. experts warn

100%.

Late Stage Capitalism is the biggest danger in AI alignment, and that is not just doomerism.

It is how you get Econ 2.0 turning all matter in the light cone into computronium or a paperclip scenario.

Those are possibilities mapped out by AI researchers now.

quote: Global Digital Compact.

Governments will take control of the internet and tech development, implement digital borders, wreck the potential of the internet, and weaponise it for their new cold and world wars.

Corporations do things badly, but you need governments to really trash everything beyond hope.

No problem to leave it to the market forces, because it can't and won't work. The market forces just will burn their fingers and bite their teeth out. Mathematicians know or let's say they easily should know, that it is impossible to function, and neither another quadrillions of tweaked model parameters, nor hardware scaling-up with faster chips nor a future quantum computing can avoid it all failing finally. Well, mathematicians are a rare minority, so the big rest including those market forces will probably need a bit more time with their obviously longer learning curve.

No problem to leave it to the market forces, because it can't and won't work.

If only that were true. If you use a computer online on any day, you are already interacting with AI. The old "rules based" responses are mostly gone already. Do you use translation tools? That's AI now. Do you use image or audio recognition tools? That's augmented with AI as well. Whenever you are shown an advertisement, AI was almost certainly used to create the images, videos (or portions of them).

If you want to avoid AI in a day, walk from your home to your garden and use simple tools to weed and plant all day.

This is a crucial wake-up call about the global governance gap in AI. The idea of a "light-touch" coordination structure sounds like a good starting point, but will it be enough to address the rapid and unpredictable changes in AI? Global cooperation is key, but without stronger measures, can we really keep pace with the risks?

Yes, I can, dear @Data. And even you don't use grammarly for a better readable output. lol