Formal verification spat: specs are inherently limited against real-world threats
davidmanheim · x · 2026-10-26
In a debate over formal verification's value, davidmanheim argues that moving from verification to formal verification is great where possible, but doesn't change the core concern: specifications are inherently limited compared to the real world and can't exhaust the plausible threats worth considering. He earlier stressed that verification's main value is that it happens at all, with formal verification proving the check follows the spec — while interlocutor akanoego insisted "spec is easier to check than code" is the whole value proposition. The exchange highlights a real AI-safety question about spec coverage.
More from AGI Musings
- Anti-AI content is algorithm gold and very profitable, quips commentator — annetgriffin · 2026-10-26
- Silicon Valley breakdown: five Stanford-area giants worth ~$18T — FinanceYF5 · 2026-10-26
- Anthropic says Claude leads 26% of its AI R&D — but self-improvement still has limits — Exponential View (Azeem Azhar) · 2026-10-26
- Humanoids shattered forecasts: half-marathon record in 2026, 100m faster than Bolt — ben_j_todd · 2026-10-26
- Stephen Casper on podcast: what if AI becomes a parasitic peer species in cyberspace — StephenLCasper · 2026-10-26
- Behind the EMNLP keynote: LMs as simulated learners to generate child language acquisition hypotheses — najoungkim · 2026-10-26