Conceptual Reasoning Fellow

Anthropic; United States; Hybrid; USD 300,000 - 500,000; Full time

Benefits offered by this job

Mentorship from Anthropic researchers

Access to Claude models and evaluation

Fellows program with extended research

One-week in-person orientation in San 

Job summary

Anthropic invites researchers and experts to join a four-month fellowship blending philosophy, cognitive science, economics, policy, and law with Claude's capabilities. You will evaluate, improve, and apply Claude in your field, mentored by Anthropic researchers, with a remote-friendly setup across the United States and a one-week orientation in San Francisco.

Applicants must hold a PhD or equivalent research experience and have hands-on Claude usage; US work authorization is required;

Qualifications

• Hands-on experience using Claude and Claude Code in your day-to-day workflow.
• PhD in philosophy, cognitive science, economics, policy, or law, or equivalent research experience.
• Authorized to work in the United States for the duration.
• Ability to work in a fast-paced, high-agency environment for a few months.
• Not currently able to sponsor visas for fellows; must already have work authorization in the United States.

Responsibilities

• Identify key open problems in your field, assess approaches, and use Claude to make progress on them.
• Identify grand societal challenges where your expertise matters and assess Claude’s strengths/ weaknesses.
• Design evaluations and RL environments to test models in your field.
• Team up with other Fellows to build environments that push Claude on judgment-heavy problems.

Skills

Claude usage

Research writing

Education

PhD in philosophy, cognitive science, economics, policy or law

Job description

About the company

the company’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.

Program Overview

Many of the questions that will shape our future have no single checkable answer. Reasoning well about them, whether in law, policy, philosophy, economics, cognitive science, or AI safety itself, takes the kind of judgment experts spend years building. We believe frontier AI, developed safely, can become a real partner on these questions, but getting there means working closely with the people who know these fields best.

This fellowship brings researchers and experts from philosophy, cognitive science, economics, policy, and law into the company for four months to work alongside our research teams. Fellows pair deep domain expertise with our frontier models and internal tooling to evaluate, improve, and apply Claude's capabilities in their field.

What to Expect:

• A concrete, scoped project matched to your background, with mentorship from the company researchers.

• Access to frontier Claude models and internal evaluation tooling.

• A cohort of incoming Fellows to collaborate with and a connection to the company's broader research community.

Compensation:

$100,000–$167,000 USD for the four-month fellowship (equivalent to $300,000–$500,000 USD annualized), and other benefits.

Key responsibilities

• Identifying key open problems in your field, assessing approaches to solving them, and using Claude to make progress on them.

• Identifying grand societal challenges which the expertise of your field is crucial for addressing, and assessing Claude’s strengths and weaknesses in tackling them.

• Designing evaluations and RL environments that test models in your field systematically, turning your expertise into working tests of how well Claude reasons in your field.

• Team up with other Fellows in your field to build environments that push Claude on the hardest, most judgment-heavy problems your discipline has to offer.

Minimum qualifications

• Are excited about the potential for AI to reason well about hard questions in their field where expert judgment matters.

• Hands-on experience using Claude and Claude Code in your day-to-day workflow.

• Hold (or are completing) a PhD in philosophy, cognitive science, economics, policy or law, or have equivalent research experience.

• Are comfortable working in a fast-paced, high-agency atmosphere for a few months and delivering concrete results.

• Existing authorization to work in the United States for the duration of the program.

Preferred qualifications

• Experience designing evaluations, benchmarks, datasets, or expert rubrics.

• Demonstrated expertise in your field, shown through research, scholarship, or expert-level analytical writing.

Representative projects

• A philosopher building hard cases that require normative judgment in ethics or formal epistemology.

• A cognitive scientist designing experimental-design and model-comparison tasks.

• An economist writing modelling and identification problems with ground truth.

• A policy scholar drafting regulatory and cost-benefit analysis tasks with expert rubrics.

• A legal scholar building AI-policy-relevant law datasets.

Logistics - Fellows

• The program runs from January 12th to May 8th 2027, and is full-time for the duration; some fellowships may be considered for an extension.

• Applications are open until October 28, 2026, and we plan to make offers by November 20th. Later stages will typically include a take-home exercise and conversations with our research team and potential mentors about your research interests and possible project directions.

• We ask that all Fellows attend a one-week in-person orientation in San Francisco.

• The default location is in San Francisco. We will consider remote engagements within the United States in certain circumstances.

• We are not currently able to sponsor visas for fellows. To participate in the Fellows program, you need to have or independently obtain full-time work authorization in the United States.

• This is a fixed-term contract position. Fellows are engaged through an employer of record, rather than employed directly by the company.

The below are the company's policies for full time roles. These do NOT apply to the Fellows Program.LogisticsMinimum education:

Minimum education: Bachelor's degree or an equivalent combination of education, training, and/or experience

Required field of study: A field relevant to the role as demonstrated through coursework, training, or professional experience

Minimum years of experience: Years of experience required will correlate with the internal job level requirements for the position

Location-based hybrid policy: Currently, we expect all staff to be in one of our offices at least 25% of the time. However, some roles may require more time in our offices.

Visa sponsorship: We do sponsor visas! However, we aren't able to successfully sponsor visas for every role and every candidate. But if we make you an offer, we will make every reasonable effort to get you a visa, and we retain an immigration lawyer to help with this.

We think AI systems like the ones we're building have enormous social and ethical implications. We think this makes representation even more important, and we strive to include a range of diverse perspectives on our team.

Your safety matters to us. To protect yourself from potential scams, remember that the company recruiters only contact you from@the company.comemail addresses. In some cases, we may partner with vetted recruiting agencies who will identify themselves as working on behalf of the company. Be cautious of emails from other domains. Legitimate the company recruiters will never ask for money, fees, or banking information before your first day. If you're ever unsure about a communication, don't click any links—visitthe company.com/careersdirectly for confirmed position openings.

How we're different

We believe that the highest-impact AI research will be big science. At the company we work as a single cohesive team on just a few large-scale research efforts. And we value impact — advancing our long-term goals of steerable, trustworthy AI — rather than work on smaller and more specific puzzles. We view AI resea