5 Comments

It's a great question for David and I apologize for jumping in, unaddressed.

It seems to me that there is no such thing as "AI alignment with human values," so long as there is no such thing as some homogeneous, monolithic notion of humanity. Perhaps AI can help humanity to finally coordinate amongst our seemingly hopelessly tribal selves, along our truly best, most peaceful, and enlightened self-interests; however, until such a time, the most simplistic expressions of "the AI alignment problem," simply do not seem to exist, from this ridiculous autistic anon's perspective, anyway.

That's probably too clunky and crude a way of putting it, so I respectfully yield back the remainder of my time and ask unanimous consent to revise and extend my remarks. ✌️

Expand full comment
5hEdited

You were once very intense about alignment, and here you're saying de-regulation is better - which strikes me as not worrying about alignment.

What has really changed that you no longer think alignment is at all worth worrying about? It seems like you're not worried about the effect upon human society of AGI disruption.

edit - 50 year leadership, it would seem, is permanent, given the acceleration of AI would likely remain exponential to our advantage.

Expand full comment

I explained it in the video

Expand full comment

Great work, as alwaysm, Sir. Big developments for most users like you said, as big as the first iPhone. OpenAI Operator is not something new, Agent Zero does that and much more. Opensource is important!

Agent Zero 🤖 Can now use BROWSER 🌐but it's much more than that.

https://github.com/frdel/agent-zero

Expand full comment

What Scott said * 1,000. 🙏

Brilliance and boneheads in all tribes.

#TeamGrandStruggle

🤖🧠⚖️💖🙏

Expand full comment