OpenAI DevDay's overlooked big idea: agents that take full responsibility like employees
jxnlco · x · 2026-10-05
Peter Gostev highlights what he calls one of OpenAI DevDay's biggest yet most ignored ideas: today's agents all assume a human is responsible and prompting them, while the real future product line is an agent that takes full ownership of a function — accounting, email marketing — like an employee would.
His argument: you hire people not just for skill but for responsibility — that's why zero-experience hires are fine, because someone is thinking about it so it's not all on you. Current agents are the opposite: they let you expand into more areas, yet bear zero responsibility, so everything new lands 100% back on you.
More from AGI Musings
- If AI Lets Anyone Execute Fast, How Do You Actually Collaborate Instead of Compete? — thejasminejade · 2026-10-05
- Gary Marcus: AI oversight needs multiple independent evaluators, like real science grant panels — GaryMarcus · 2026-10-05
- Study Links AI-Writing Traces in Dissertations to Leaving Academia — paulnovosad · 2026-10-05
- Anthropic's Jack Clark shares the best take from The Curve: 'we can still heist the Mona Lisa after the singularity' — jackclarkSF · 2026-10-05
- Economist stops reviewing AI-written papers: instant generation breaks peer review reciprocity — paulnovosad · 2026-10-05
- P(doom) Is a Bayesian Prior, Not Frequentist Probability — AI Community Reopens the Debate — irinarish · 2026-10-05