we need more advanced evals to push forward the field!
@alexandr_wang
-
LastExam: Paper, Dataset, and Code Release
By
–
Paper, dataset, and code: http://
lastexam.ai
NYT article: -
American Energy Strategy to Power Frontier AI Systems
By
–
5/
Unleash American energy to support AI boom. Trump understands the required energy needed right now to fuel frontier AI systems. The newly created National Energy Council should launch an aggressive energy plan to produce the lowest-cost energy and electricity globally. -
Federal Agencies Must Become AI-Ready by 2027
By
–
4/
Make all federal agencies AI-ready Government data is not optimized to support AI, which means the benefits on things like defense or healthcare cannot be fully realized. Executive action on AI should require our agencies to become fully AI-ready by 2027. -
Building America’s Future Workforce Through AI Job Creation
By
–
3/
Build the workforce of the future in America Great AI creates jobs. AI engineers, expert data annotators, AI agent trainers AI will create up to 50 million new jobs and provide up to $15.7 trillion of global economic output by 2030. We need to establish an interagency task -
Reallocation of US AI Investments: Prioritizing Compute and Data
By
–
2/
Allocate America’s AI investments in the right places. Our government spends ~90% of investments on algorithms, neglecting compute and data. The leading global tech cos have been investing ~60% on compute, 30% on data, and 10% on algorithms. Our government must adapt our -
New Administration’s AI Strategy to Compete with China
By
–
1/ New Administration, same goal: Win on AI Our ad in the Washington Post, January 21, 2025 After spending the weekend in DC, I’m certain this Administration has the AI muscle to keep us ahead of China. Five recommendations for the new administration
-
Scale AI Formalizes Merit-Based Hiring Policy
By
–
Today we’ve formalized an important hiring policy at Scale. We hire for MEI: merit, excellence, and intelligence. This is the email I’ve shared with our @scale_AI team. ——————————————————— MERITOCRACY AT SCALE In the wake of our fundraise, I’ve been getting a lot of questions
-
DeepSeek V3: China’s AI Model Rivals GPT-4o with Less Compute
By
–
It is quite fitting that DeepSeek, China’s leading LLM lab, releases its latest model V3 on Christmas.
– on-par with GPT-4o & Claude 3.5 Sonnet
– trained w/10x less compute The bitter lesson of Chinese tech: they work while America rests, and catch up cheaper, faster & stronger -
Veo 2 Generates Image of Cheese Dragon Dressed as Santa Claus
By
–
Veo 2 is so good.
— Alexandr Wang (@alexandr_wang) 24 décembre 2024
"A dragon made of cheese dressed as santa singing an opera with a t-rex" pic.twitter.com/zz4UIS7V1JVeo 2 is so good. "A dragon made of cheese dressed as santa singing an opera with a t-rex"