On Thursday, the Lord Institute announced its first batch of Slingshot grants aimed at “advancing the science and practice of artificial intelligence.”
Designed as an accelerator for researchers, the Slingshots program aims to provide resources not available in most academic environments, including funding, computing power, and product and engineering support. In exchange, the recipient promises to create some kind of final product, whether it’s a startup, an open source codebase, or another type of work.
The initial cohort is 15 projects, with a particular focus on the difficult problem of AI evaluation. Some of these projects are familiar to TechCrunch readers, such as the command-line coding benchmark Terminal Bench and the latest version of the long-running ARC-AGI project.
Others take new approaches to long-established assessment problems. Formula Code, built by researchers at the California Institute of Technology and the University of Texas at Austin, aims to assess an AI agent’s ability to optimize existing code, while Columbia-based BizBench proposes a comprehensive benchmark for “white-collar AI agents.” Other grants explore new structures for reinforcement learning or model compression.
John Boda Yang, co-founder of SWE-Bench, is also part of the group as the leader of the new CodeClash project. Inspired by the success of SWE-Bench, CodeClash evaluates code through a dynamic competition-based framework. Yang hopes so.
“I think it will drive progress if people continue to evaluate on core third-party benchmarks,” Yang told TechCrunch. “I’m a little worried about a future where benchmarks become company-specific.”
