This is my favorite Dario Amodei essay since "Machines of Loving Grace". - It starts with a discussion of the potential benefits of AI and then acknowledges the risks. This is exactly the right way to present the technology to the public. - "We must slow the pace at which we improve the capabilities of AI models. Progress will still seem fast, and we must make wise use of the time we gain." - I think the proposal to embed third-party evaluators within Anthropic (initially voluntarily) is great, and other frontier labs should strongly consider doing the same. - This is, to my knowledge, the very first proposal that expressly encourages that "pacing the frontier" cannot result in us ceding our lead on AI to China: "Pacing within democracies will be limited by the lead that US companies have over authoritarian regimes, chiefly the [CCP]. If we slow down by more than this amount, then (unpaced) CCP-associated projects will pull ahead, creating significant national security risk... a key part of pacing within democracies is to keep democracies’ AI lead over autocracies as large as possible, to give us the breathing room we need in order to pace effectively." - It makes the call to the USG to allow for a way for the frontier labs to cooperate on safety without violating the antitrust laws - very sensible and definitely needed.