What 4,523 AI investment forecasts revealed about model disagreementipulseai.substack.com3 points·by russlan·vor 5 Tagen·0 comments
russlan·vor 21 Tagen·discussimo model switching only works if you normalize outputs and keep model specific weirdness behind well designed adapters
russlan·vor 21 Tagen·discussgeneric model benchmarks are less useful than workflow benchmarks.Nowadays it's all about workflows , think prompt injections, tool permissions is a big one, logs and lately what anthropic claims is the model distillation risks
Interested in AI agents, evals, product engineering, fintech/market data, and making model outputs easier to audit.