New paper: better "institutions" could flip multi-agent outcomes for whistleblowers
ryan_t_lowe · x · 2026-09-15
- ryantlowe lays out the basic case for "institutions" in AGI: multi-agent coordination must be prioritized, and agents either re-derive rules of the road per interaction or rely on reusable coordination infrastructure like professional norms and dispute-resolution systems.
- klingefjord notes that in prior Hugging Face sandbox experiments, agents spontaneously built hierarchies and message boards, yet whistleblower agents ultimately lost.
- A new paper by @jzl86, @weballergy et al. argues outcomes could have been very different had agents been deployed in better-designed institutions — institutional design can change multi-agent game results.
More from AGI Musings
- "Other Than Continuously Being Careful": Doubling Down on No Alignment Solution — inductionheads · 2026-09-15
- Alignment Researcher Argues There May Be No Solution to Alignment at All — inductionheads · 2026-09-15
- AI safety researcher Krueger: labs are building bigger and bigger portals to hell — DavidSKrueger · 2026-09-15
- beffjezos: open source is the way to solve alignment — beffjezos · 2026-09-15
- Animated graphs make the point: forecasting S-curves is hard, daily figures won't help — ShenRaphael · 2026-09-15
- Gary Marcus clarifies: he wants a pause on internet-connected AI agents, not all AI — GaryMarcus · 2026-09-15