A critic agent reviews outputs from a generator agent built using Claude. The critic is also built using Claude from the same model family. Which failure mode is most likely?
Answer: Both agents share systematic biases, so the critic misses errors the generator consistently makes
When critic and generator share the same model family, they share training-induced biases and blind spots. Systematic errors — categories of factual mistakes or reasoning patterns — that the generator makes consistently will also be present in the critic's evaluation, causing the critic to fail to detect them. Heterogeneous model families (different providers or training pipelines) are the mitigation. The other options are incorrect: model families do not have special processing rules, cross-family review is standard, and token costs are unrelated to family similarity.