Models that are good at math benchmarks tend to be good at coding and reasoning benchmarks too, pointing to a common factor driving AI capabilities.
We find that AI benchmark scores are nearly as correlated across domains (0.68) as within them (0.79).
Fields medalist thinks that AI might surpass mathematicians in 10 years!
I am very proud of the work we are doing and incredibly grateful for the support we have received from our audience. Thank you for considering supporting Epoch AI!
epoch.ai/support-us
More in-depth feedback is even better. For example, I would be interested in learning which of our outputs our donors find most impactful, and of decisions informed by our work.
Even short feedback like “I am a software engineer at Google, and I donated $200 to Epoch AI because I read your 2030 scaling report and it clarified that power infrastructure won't impede AI scaling” will help us understand our audience and prioritize future work.
However, we are starved for feedback on how our work impacts the world. A useful contribution would be to donate $100-$1000 to Epoch AI and then email us explaining why you chose to support us.
The impact of small individual donations (<$1000) is limited. Our yearly spending is close to $3M, and I expect we will raise $10k to $100k from small donors this year, which would be enough to fund an extra data insight or a short report.
Our room for funding right now is quite high – we can productively absorb an additional $10M over the next two years. I am actively looking for partners who can provide us with over $1M in donations and feedback.
First, Epoch AI accepts donations from individuals, and are tax-deductible from most countries. Your donations will be used to further our mission of improving the public understanding of AI's trajectory.
epoch.ai/support-us
A @epochai.bsky.social follower asked me about our funding situation and whether we accept donations from individuals. I thought I would write a few words for potential donors.🧵
Could you have trained GPT-4 on 2012 GPUs?
Today, we’re releasing an interactive distributed training simulator that allows you to answer this question, among many others!
We've just launched our AI Benchmarking Hub!
This is a new platform for rigorous, independent evaluations of AI model capabilities, featuring interactive visualizations and in-depth analysis. (1/8)
epoch.ai/blog/introdu...
We are taking on a mission to track progress in AI capabilities over time.
Very proud of our team!
