Peter McCrory

This is an important limitation of the model. We focus on the supply side of the AI transition (what AI can do, how fast it spreads, how quickly workers change jobs). Prices are flexible and aggregate demand is equal to what the economy can produce. More thoughts in 🧵

Broadly agree. Some practical implications: (1) Prioritize analyses that can be periodically updated with new data (2) Do your work in the open (revise your views based on new evidence) (3) Recognize uncertainty; make falsifiable predictions (4) Be earnest & humble

New research: Project Swap To see what works & what breaks when agents are sent into a marketplace, we made a mini barter economy of Claudes. It gave us an early look at the economy of the future––and the challenges to overcome to make agent-run markets efficient, safe & fair.

I've learned a lot from @jasonfurman over the years. So, it was a real privilege to join him on stage for a great conversation discussing the economic implications of AI. I appreciated the audience's patience with us as it got a bit wonky at points. Fortunately, Jason told me afterward that I only used the phrase "infinitely elastic" once. While Jason's questions were great, I was particularly impressed by and thankful for the questions that the students asked during and after the event. Ensuring that the benefits of AI are broad-based and risks are appropriately mitigated requires that we ask hard questions — with AI, there are many.

Please read this essay. The transformative potential of AI to bring immense benefits to the world requires that we appropriately manage the risks that rapid progress brings. Embedded evaluators is a sensible first step. We have agency. Let’s exercise it.

Nice, complementary research to what we shared today. Evaluating the key economic forces that govern how large/small an effect AI will have on growth in the years ahead (and assessing which path we may be on today) is critical work. Great work @alexolegimas @ben_moll

This is an nice report tackling one of the most important questions: How might AI affect science and innovation? What impact is it already having today? Great work, Mihai and team.

Great discussion w/ @jackclarkSF on our econ scenario research. The goal is not to make a forecast per se, but to understand the range of possible outcomes and the conditions under which they may emerge. The hope is to clarify sources of disagreement about our uncertain future.

What work can robots do today? And what might that imply for the structure of the economy in the years ahead? New research from @rclegateyang and Maxim Massenkoff out today exploring these questions.

Great to see Anthropic's new mission report today, especially the new mission pillar focused on empowering every person pursue to pursue a good life, including meaningful work. In earlier research, that's one of the main things that people hope for in an AI-transformed economy — working toward this vision is a big reason why I joined Anthropic in the first place.

Good clarification from @ben_moll . These are model-implied scenarios for growth rather than direct GDP forecasts. Anthropic staff expect more capable models, faster diffusion, and greater productivity gains from AI when its used. In the model, such forces accelerate growth.

Worth your time and consideration.