Does the "outperforming" conclusion incorporate the appropriateness of decisions? Or just if things are technically correct. Without human eyes on cases, things could easily get very off track. AI can do a lot of data wrangling, but there is no conscience.