About the Role

Turing is looking for an AI Research & Evaluation Specialist to help build a new evaluation benchmark for frontier AI browsing agents. This role is not about writing articles, creating generic research questions, or being an expert in one particular subject.

Instead, you’ll be doing investigative research, your job will be to create research problems that are genuinely difficult for advanced AI systems to solve even when those systems have full access to the web and multiple opportunities to search for the answer. You’ll begin with a fact that can be independently verified. From there, you’ll work backwards to design a natural-sounding question that makes that fact extremely difficult to discover.

What You’ll Produce

As part of the role, you’ll create research challenges that include:

• A natural-language research question with a short, stable, and objectively verifiable answer
• Multiple clues that can each be independently checked
• Specific constraints covering different types of information, including dates, people, places, organizations, works, events, records, and quantities
• A detailed evidence trail showing exactly how the answer can be verified
• A validation record documenting the obvious searches you performed and what those searches returned
• Precise citations pointing to the relevant pages, tables, sections, records, or documents

The goal is to create questions that are difficult to solve but straightforward to verify once the correct evidence has been found.

What We’re Looking For

You should be comfortable investigating unfamiliar topics from scratch and following information across different sources. You’ll ideally have demonstrated experience with open-web research, including finding primary records and working with sources such as government databases, institutional archives, registries, libraries, and PDFs.

Strong attention to detail is essential.

We’re looking for someone who doesn’t simply provide a link and say, “the answer is here.” You should be able to identify the exact page, table, section, record, or passage that supports a claim.

Apply Here