New essay: “Which Future?”
michaelnotebook.com/whichfuture/...
On how to wisely navigate risks from transformative technology, especially artificial superintelligence (ASI). This condenses much of the key thinking from my more extended essays
This was terrific:
Thanks. Bsky didn't seem to be going in a great direction, so I only check briefly once or twice a month. I was quite irritated by this comment: bsky.app/profile/bnew...
I enjoyed this thoughtful account of how Open Phil approach progress: support technical and scientific advances, but keeping a thoughtful eye on safety, too, viewing it as part of progress, not competitive with
Actually, it might be in the Barenco et al 1995 paper (the one with 9 or so authors, on gate universality). I suspect that's a good spot to check!
Wrote it down? I'm not sure. IIRC I learned it from Manny Knill in 1996, but my memory is very fuzzy
I surprised myself by writing a blog post about religion this morning 🤭
statmodeling.stat.columbia.edu/2025/08/01/a...
I thank (blame?) @economeager.bsky.social and @michaelnielsen.bsky.social for inspiring me with their recent essays
Reading this with much interesst!
Reminds me of this paragraph from @michaelnielsen.bsky.social about human willingness to delegate to world-ending authority automated systems (whether because oversight is impossible or just inconvenient)
michaelnotebook.com/xriskbrief/i...
"Maybe if we add another rule / consultation that will help fix the decay" is kinda an attempt at distilling the inverse philosophy...
Good choice
(I just listened to Serkis read "The Silmarillion" and "The Lord of the Rings", and it was great!)
A few thoughtfully-chosen excerpts from my "How to be a wise optimist..." essay:
Lovely thread:
Thanks. Ugh
Do you know what happened with this? I missed that Cremieux was invited (and would likely have declined to participate had I known)
Yeah, I had people like Ostrom, Mancur Olson, Schelling, Axelrod, and even Fukuyama in mind - really, the whole game-theoretic foundations for the evolution of co-operation
At some point I realized that everything online breaks, so what you want is handwritten static files hosted in the most stable place you can think of. GitHub Pages will eventually die, but I think that site is safe for a while...
Thanks. I'm certainly never going to be known for my design skills...
Or, um,
ChatGPT, took 10 seconds for me to type the prompt
A thoughtful post on AI Alignment:
Congratulations!
You might well be able to write quite a funny paper about "Emergent stupidity" ("we trained it to be anti-woke, and everything else got worse too...")
This is a really interesting speculative thread:
I mean two people who think they're working on homework at school, but are actually talking about shampoo. One of them just expressed surprise that their work is taking so long. I refrained from commenting "That's because you've only worked 10 mins over 2 hours..."
I find it oddly irritating when people think they're working but are actually hanging out. Like... just own that you're socializing, and do it well, rather than working totally ineptly...
