NewsMore in News→
Campbell Brown’s Forum AI is betting expert-built benchmarks can clean up high-stakes model answers
Key Takeaways
- Forum AI evaluates foundation-model performance on high-stakes subjects such as geopolitics and finance.
- The company says it recruits experts to build benchmarks and trains AI judges to reach about 90% agreement with them.
- Brown argues major model developers emphasize coding and math more than information accuracy.
- She says current models still show sourcing problems, political bias, and missing context on sensitive topics.
DE
DT Editorial Team··via techcrunch.com