The ARC Prize Foundation, a nonprofit organization focused on advancing artificial general intelligence (AGI) through benchmarks, has launched ARC-AGI-2, a new tool designed to challenge AI reasoning capabilities. The foundation also previewed plans for ARC-AGI-3, which will address future AGI challenges. Public donations have been opened to fund the development of ARC-AGI-3. The foundation has received financial contributions, including $50,000 donations from Eric Schmidt, Tyler Cowen, Dharmesh Shah, and Tim Shi, as well as $20,000 from Zico Kolter, $25,000 from Aaron Levie, $30,000 from Vercel, and $5,000 from Logan K. Additional support has come from a16z and Martin Casado. The funding goal is $2.5 million, with over 54% of the target already achieved. ARC Prize's mission is to identify and quantify capability gaps between humans and AI, providing benchmarks that inform research and policy decisions. The organization has also announced a collaboration with Vercel and highlighted Logan K's statement about the importance of evaluations as AGI progresses.
.@OfficialLoganK just donated $5K! From Logan: "As we get closer and closer to AGI, evals are becoming more difficult, yet more important. I'm excited to personally support the ARC Prize Foundation so we can better know where we are on the path to AGI." https://t.co/ML0pukwNKN
When I learned about the ARC Prize ~9 months ago, I was immediately intrigued. It taught me something about what LLMs can NOT do. Given how amazing LLMs are at so many things (and how fast they’re improving), that was an important lesson. I’m thrilled to contribute to @arcprize… https://t.co/Muu1B2kngy
When I learned about the ARC Prize ~9 months ago, I was immediately intrigued. It taught me something about what LLMs can NOT. Given how amazing LLMs are at so many things (and how fast they’re improving), that was an important lesson. I’m thrilled to contribute to @arcprize… https://t.co/Muu1B2kngy