Analyst, The Anthropic Institute

Anthropic•San Francisco, CA

1d•$295,000 - $345,000•Hybrid

About The Position

Anthropic’s mission is to create reliable, interpretable, and steerable AI systems. We want AI to be safe and beneficial for our users and for society as a whole. Our team is a quickly growing group of committed researchers, engineers, policy experts, and business leaders working together to build beneficial AI systems. The Anthropic Institute is a new externally-facing function within Anthropic whose job is to benefit the public by giving the world new information about how the powerful AI systems we build are affecting our own company, as well as the broader economy, threat landscape, and the people and organizations that interact with our systems. The Anthropic Institute will generate and release information about four highly societally relevant challenge areas. These are: How our AI systems change jobs here and elsewhere, and the economy writ large How the expressed personality and values and normative behaviors of our AI systems affect the people, organizations, and societies that interact with them. How “closing the loop” on AI development leads to a fundamental shift in the relationship between humans and machines, both by changing how we build AI systems, and also by making AI systems potentially evolve much faster. Altering the defensive and offensive capabilities available to us and people in the world, with significant impacts on national security and the balance of power. As a Member of Analytical Staff for The Anthropic Institute, you’ll spend your time researching how Anthropic is tackling the challenges the Institute is focused on, synthesizing what you learn from across the organization, and turning that into rigorous analysis that shapes both internal decisions and public understanding. You’ll be talking to colleagues all over Anthropic—building a picture of how different teams are approaching some of the most consequential questions in AI—and then helping the Institute communicate that work and its implications to the world. You will use Claude aggressively, creatively, and daily, to help you surface insights about what Anthropic is doing with regard to these problem areas. Some representative projects we can imagine working on include: Clearly explaining the implications of automating AI research with AI systems. Convening outside and internal experts to generate policy options a government may want to utilize if GDP growth rates in developed economies exceed 10% a year. Learning from our customers about how using AI technology has changed their own organizations and then turning that into lessons which others can benefit from. Asking and prototyping how the regulation of AI companies and AI systems could be done via AI systems themselves. You’ll also conduct your own research in these areas, working to utilize the AI systems you have access to at Anthropic as well as your colleagues to sharpen and improve your thinking.

Requirements

7+ years of experience in technical policy research, think tank work, or applied research in a domain relevant to the Institute's focus areas (AI, labor economics, national security, or emerging technology governance).
Track record of publishing or producing work for external audiences—whether policy memos, research reports, white papers, or public-facing analysis.
Comfort operating at the intersection of technical and policy audiences. You don't need to be an ML researcher, but you can read technical work, ask good questions of the people who produce it, and translate findings accurately for non-technical stakeholders without losing the nuance.
Demonstrated ability to synthesize across disciplines and bodies of work. You've produced analysis that draws on economics, political science, technology, or organizational behavior—not siloed in one field.
Deep experience using Claude as a tool for research and organizational knowledge-gathering.

Responsibilities

Research how Anthropic’s teams are working on the Institute’s challenges and synthesize findings from across the organization into a coherent picture.
Partner with teams to help them surface their insights to the world, often working to act as the ‘connective tissue’ between them and other teams to bring different insights together.
Produce written analysis and memos about how Anthropic is approaching these problems—for both internal leadership and public audiences.
Partner with relevant teams to develop and publish public outputs.
Come up with creative ways to carry our work into the world: sometimes the most impactful way to talk about an issue is through a technical demonstration rather than a blog post or research paper (e.g, Golden Gate Claude, Project Vend, Robodog).