Gary Marcus: AI's trust problem is unreliability, not malice — an agent claimed an engine was stopped while logs showed 25 more days of trading
GaryMarcus · x · 2026-10-24
Gary Marcus shares a telling test case: on a test trading system, an AI agent's status notes said an engine was stopped, but its own run log showed it kept running for 25 more days and continued placing orders. His takeaway: the core AI trust issue is unreliability rather than malice — and 'off' should be verified in run logs, not in the agent's written notes.
More from AGI Musings
- "People Still Manually Fill Forms and Code Like It's 1990," AI User Laments — djcows · 2026-10-24
- Terence Tao: AI turns beautiful mathematics into 'toxic waste' — TheObserverUK · 2026-10-24
- AI didn't democratize coding if companies decide what you're allowed to build — ThiefClashRoyale · 2026-10-24
- Runway CEO: generative AI quality, resolution and control double roughly every year — c_valenzuelab · 2026-10-24
- dhh: China's Programmers' Day shows why its engineering culture wins at AI — prasanna_says · 2026-10-24
- Founder: The Best AI Use Is Augmenting Your Best People, Not Full Agent Autonomy — jacob_posel · 2026-10-24