Richard Ngo - Reframing AGI Threat Models [Alignment Workshop]

In “Reframing AGI Threat Models,” Richard Ngo suggests defining ‘misaligned coalitions’-groups of humans and AIs that might grab power in illegitimate ways, from terrorist groups and rogue states to corporate conspiracies. This alternative framework shifts focus to the nature of coalitions and their risk potential, whether from decentralization or centralization.
Highlights:
🔹Misuse vs Misalignment - Distinction not useful on a technical or governance level
🔹Misaligned Coalition - Humans + Als attempting to seize power in illegitimate ways
🔹Small-scale actors - Risks from decentralization support offense, compute controls, and international regulation
🔹 Large-scale actors - Risks from centralization support open source, transparency, and limited government powers
The Alignment Workshop is a series of events convening top ML researchers from industry and academia, along with experts in the government and nonprofit sectors, to discuss and debate topics related to AI alignment. The goal is to enable researchers and policymakers to better understand potential risks from advanced AI, and strategies for solving them.
If you are interested in attending future workshops, please fill out the following expression of interest form to get notified about future events: far.ai/futures...
Find more talks on this KZbin channel, and at www.alignment-...
#AlignmentWorkshop