Die Shrink 12: The Future of Work
🕑 Added 2020-01-31 10:01:00 +0000 UTCWhat skills will matter in the future? Is there a marketing bubble? Are people prepared for the future of work?
In many ways it feels like we are entering a new era of Automation that may require a new type of "Renaissance Man." We discuss all of this and touch on AI and Automation heavily.
Comments
The Immortal Cameraman
Welp, I'm going to be in the fitness industry soon. That's not going away anytime soon. 👌
Angmar
Funny you mention Elevator workers, I ran into some poor, bored, souls sitting in elevators pressing the button for your floor in elevators in the Mall of Asia in the Philippines
Edward Huff
In terms of timelines, I have no idea. I just feel like the error bars are large. Could be 100 years. Could be 10. I have no idea. People like Kurzweil who predict a specific time and date are being silly. Save for George Hotz who predicts 2038, purely because that's the unix time rollover. What I do think is, is it really does look like programming a super-human AI that maximizes a human-friendly utility function will be much, much harder than a super-human AI with an arbitrary utility function. And I think, as a species, we are smart/stupid enough to stumble into the later eventually. Maybe we just need a Butlerian jihad.
Moore's Law Is Dead
So if you were to ask us 2-3 years ago, me and Dan would have said "Yeah AI is a huge threat." But as time goes on our mind is changing. It's not that I think it can't become a threat, I just doubt it's as pressing as people think...and I think people are in fact overestimating human intelligence by assuming we can just duplicate our intelligence ;).
Edward Huff
Fun conversation about a fascinating topic. I think the case for worrying about AI is much more solid than you guys do. This great paper by Stephen M. Omohundro is worth the read if you are interested in a contrary take: https://selfawaresystems.files.wordpress.com/2008/01/ai_drives_final.pdf. The AI risk people are very smart. They have thought deeply about the topic. Superficial ideas like "just keep it in a box" or "intelligence is not well defined" all fail under scrutiny. Anyway, Omohundro's paper is the shortest, clearest explanation I have seen of the basic problem of AI risk.