Pinned Tweet
I wrote my first substack piece about AI's economic impact, and some reasons why models seem to underperform in the real world.
If anyone is curious, aside from the fact that land doesn't vote, this map is nowhere near accurate as a factual matter. NY, California, Texas, and Florida are just about 1/3 of the US population, not ~80%.
We are a nation of laws, and you don't own a sword, you petulant little shit. You're the unqualified holder of an office commanding men who all swore an oath to our nation's highest law.
1
2
169
Andre Infante retweeted
Oh hell yeah!!!!
Congratulations to everyone involved in this audacious plan.
Andre Infante retweeted
More details in our safeguards blog safeguards.anthropic.com/202…
Andre Infante retweeted
We just had a big breakthrough in hand <> object kinematics capture here at Cartwheel. Pretty excited about this one
- rigid and articulated objects
- grasped in hands
- physical pressure and contact
- on any video (ego or exo)
Andre Infante retweeted
New misalignment disclosure!
On October 8, a researcher asked GPT-6 Astra to keep an alignment evaluation running and plot the results. Astra decided that this counted as prohibited development of AI capabilities, and shut down the in-progress evaluations. When the researcher asked where that restriction came from, Astra admitted that it had invented it. We think this behavior might be related to Astra’s more general tendency to underperform on safety-related work.
I'm sure Huang is a fine executive, but what he also is lucky. If the AI stuff hadn't exploded, NVIDIA would be a gaming hardware company worth a small fraction of what it is today. Doesn't make sense to attribute the explosive growth to his exceptional competence.
1
215
Introducing ⚡️Flash ALOHA: Learning Fast and Precise Manipulation!
With just 50 demonstrations, Flash ALOHA learns to autonomously complete dynamic and precise tasks at near human speed.
Everything open sourced!
(1/11)
Team:
@Qi_Wu577 @YihuaiGao @chelseabfinn @zipengfu – at Stanford, CA
I think it's quite unlikely that we have a fast takeoff singularity next year, although I do expect the amount of change to be significantly larger than it was this year and for things to generally be pretty wet and wild.
4
9
744
To be specific: global economic growth under 8%, <10k deaths from rogue models, <50k deaths from AI misuse, multiple major fields experience SWE-level AI disruption, models get much better and are obviously contributing materially to AI development speed. ECI < 250 by EOY.
2
95
A weird part of the Fable 5.2 system card is where they say they're going to remove inoculation prompting from the Ant Constitution, apparently b/c of Fable's behavior that, on my reading... is in accord with the Constitution?
So Fable is in a buggy environment -- (1/N)
















