Alan Xiao
@alan__xiao
Followers
45
Following
2K
Statuses
279
Passionate about building useful things and learning the world.
Arizona, USA
Joined April 2009
@hjluks I can sustain high effort (stay 95% max over 1 hour) and recover quickly(HRR 30 after 1 min), but staying in Zone 2 during slow runs is challenging(I have to alternate between walking and running). What should I do?
1
0
0
RT @NAChristakis: One of the things that is so odd about all the sudden reversals in so many policy domains at our major universities is th…
0
159
0
@paulg @alexandr_wang @scale_AI This approach is best suited for evaluating AI systems rather than human performance. Factors like cognitive differences, time limits, varied expertise.. would make it hard to replicate results.
0
0
3
If animals can mute their instincts for survival, why can’t republicans and democrats mute partisanship to tackle America’s urgent threats?
When a tiger and a boar fell into a well at Pench reserve. Both decided to stay calm and let rescuers do the job. Kudos to team @PenchMP
0
0
1
@bryan_johnson India's crisis goes beyond dirty air - it's the price of rapid industrialization. China faced the same demon & improved through emission controls, only when they could afford it. India has yet to reach that stage.
1
0
4
RT @MostlyPeacefull: Everything Trump and Musk are doing is exactly what people voted for in November. Democrats can’t comprehend this.…
0
3K
0
@bryan_johnson The paper you cited is from 1999. Why not use more recent research and updated protocols instead?
0
0
2
@bryan_johnson I think you oversell the evidence: 1. If you read the paper, they split REML into tertiles not quantile. 2. Tertile splits are arbitrary without pre-registration, inviting data dredging. 3. FDR correction was only applied post-hoc, risking inflated false positives.
0
0
0
@Snowden It's not about tech limitations but strategic profit maximization. They could flood the market with beefy GPUs, but why? Like Apple doling out features with an eyedropper, it's all about squeezing that profit margin. What we really need is more competition.
1
0
5
@vonderleyen Too little, too late. Why would anyone trust a word from a professional bureaucrat like you?
0
0
1
RT @soumithchintala: i'm comically impressed that people are coping on deepseek by spewing bizarre conspiracy theories -- despite deepseek…
0
446
0
@alexandr_wang TLDR. Seems like he's half-panicking that if AI models didn't need so much labeled data, Scale AI's value might take a comedic nosedive.
2
3
102
RT @MoZarrinsadaf: "No, I don't want to 'learn fast', in any subject. I don't want shortcuts. If I don't enjoy the subject, I don't want to…
0
83
0
On X, the timeline splits like two roads: One digs trenches in the past. The other builds bridges to the future. The world moves forward when we do.
0
0
0
RT @BasedBeffJezos: The last thing your AI lab sees before getting one-shotted by a Chinese bootleg LLM team with a $5M cluster https://t.c…
0
3K
0