My thoughts on Elon Musk’s AI warnings:

I’m not sure if it’s just the cultural echo chamber, but I feel like I’ve had or come across more “Strong AI worries” discussions in the past week than in the past few years combined.

My first reaction has been “psh, that’s crazy talk.” But then, it might not be:

  • Elon Musk has a good track record of making contrarian claims and ultimately being right.
  • I put a lot more stock in predictions for things happening within the next five years than > five years. You can mentally map out the things that need to happen for a five year prediction, while longer time horizons require gaps that might ultimately never be filled.
  • Stephen Hawking and Elon musk are both very smart.
  • The second derivative of AI and robotics feels really large. I’m right on the edge of several research communities.When I look at space, mechanical engineering, material science, and physics, it all feels slow, incremental and a bit sad. Robotics and CS just consistently makes me go “holy shit” and feel like a pathetic Neanderthal. Sure 90% of it sucks, but 90% of everything sucks and the 10% that doesn’t is mind boggling. And the thing is that more than anywhere else, AI advances compound.
  • Things that have money poured on them that don’t try to violate fundamental physics or change human nature tend to succeed. And boy does the AI/robotics fire have a lot of money poured on it right now.

Is this something we’re going to look back 30 years from now and laugh “ha, remember back in the twentyteens when everybody was worried about strong AI?” Maybe.

What do you think?

Originally written for a friend, worth having on record.