AITechTalks7 months ago
Why AI benchmarks are broken
1 min read
OpenAI released GPT-5.2 after Google's Gemini 3 Pro launch. GPT-5.2 sets a record on ARC-AGI-2 but lags behind Gemini 3 Pro on the Artificial Analysis Index and Epoch Capabilities Index. GPT-5.2 Pro ranks 8th on Simple Bench. Labs now use benchmarks for bragging rights, leading to overfitting.
Level
Hype check
Tap to vote and see what everyone thinks.
