Japanese developer argues Jev automation thresholds should follow probability and failure cost
After reading the official Jev docs and several Japanese write-ups, the author argues that rather than deciding which judgments to hand to Jev, teams should set automation thresholds by probability and failure cost, since high confidence does not mean it is safe to automate.
私なら「どの判断をJevに任せるか」より、確率と失敗コストで自動実行の閾値を決めて、どこまで人間そのものをループから外せるか、かな。 confidenceが高い＝自動化していい、でもないので。 勿論、読んだ👇 https://typesafe.ai/ https://docs.typesafe.ai/ https://console.typesafe.ai/ https://evals.typesafe.ai/ https://forbesjapan.com/articles/detail/105325 https://qiita.com/y-morimatsu/items/b396064bf6f3d84ca200 https://note.com/kagawatomo/n/n5425654d0f5d https://zenn.dev/serada/articles/20260924-jev-system-one-intro https://recruit.group.gmo/engineer/jisedai/blog/jev-system-one/ https://note.com/kiyo_ai_note/n/n27a08ff0034b https://fixit.co.jp/tips/jev-typesafe-ai/ https://www.mindstudio.ai/blog/jev-system-one-model-launch
This post quotes another post: x.com/mmchh000/status/2105825089283563660