Pinned post
I've written about race, genetic ancestry, analyses of large biobanks, and human history
gusevlab.org/projects/hsq/#…
I'll summarize the key points here 🧵:
gusevlab.org/projects/hsq/#…
I'll summarize the key points here 🧵:
Sasha Gusev
Verified account This account is verified. Learn more@SashaGusevPosts
Statistical geneticist | Associate Prof at @DanaFarber / @harvardmed / @DFCIPopSci | Blogging at theinfinitesimal.substack.com
2.9K Following 26K Followers 24K Posts
Latest posts 5
Beyond capabilities, AI2027 has gotten politics and economics badly wrong so far:
- Stock market gains
- Chinese centralization
- US government ignoring AI risk
- Stock market gains
- Chinese centralization
- US government ignoring AI risk
I feel like everyone glorifies AI 2027 and cherry picks vague data interpreted generously and ignores clear statements that were made that are simply false.View quoted post on X
For example, they predicted the stock market goes up 30% in 2026, led by OpenBrain and Nvidia. The stock market is up 7% YTD and NVDA is down 4% ytd.
The least AI-exposed company in the Mag 7, AAPL, is the best performing of Mag 7.
Sympathetic to this argument -- the models are very good -- but I think it's wrong. If you work on a complex project for multiple weeks you will start to see seams, as more and more of your time is spent repeatedly catching and explaining the same failure modes.
If you simply froze AI capability at Mythos / Astra levels and focused on decreasing the cost of compute you would totally revolutionize software development with zero existential riskView quoted post on X
Amodei: "Adults are 2-3x more worthy of charity than children, this is a number I just invented but it feels right idk"
*public is aghast*
EAs: "Oh so you hate cost-benefit analysis? We should just make decisions on vibes instead??"
*public is aghast*
EAs: "Oh so you hate cost-benefit analysis? We should just make decisions on vibes instead??"
A simple polling average beat every model this year.
This is a wonks-only postmortem, so please forgive me for one more: I think @NateSilver538 used to say that he'd prefer you just use Classic in a competition.
But if you allow it (and we should!), then the best Brier Score model of the cycle is clearly Silver Bulletin's Deluxe.View quoted post on X
