Bindu Reddy Announces New AI Benchmark
Abacus.AI co-founder and CEO says current tests miss real-world long-running loops.
TLDR
Bindu Reddy, co-founder and CEO of Abacus.AI, stated on X that all AI benchmarks are broken or saturated. She said they fail to test real-world long-running loops and that benchmarks require a revamp once models improve. Reddy announced that Abacus.AI will launch a new benchmark next week designed to fix the issue. The post presents the move as a direct response to limitations in existing evaluations.
Combined views
16.1K
1 Source, first seen 27d ago