{"id":2,"date":"2012-05-28T19:14:51","date_gmt":"2012-05-28T19:14:51","guid":{"rendered":"http:\/\/www.twonewthings.com\/gabrielrecchia\/?page_id=2"},"modified":"2025-04-01T04:58:01","modified_gmt":"2025-04-01T04:58:01","slug":"sample-page","status":"publish","type":"page","link":"http:\/\/www.twonewthings.com\/gabrielrecchia\/","title":{"rendered":""},"content":{"rendered":"
\"The<\/a>\n
\n

Gabriel Recchia<\/h1>\n

Director, Modulo Research<\/h2>\n<\/div>\n<\/div>\n

I’m a cognitive scientist working on the evaluation and alignment of large language models as the director of Modulo Research. We recently released a dataset<\/a> of expert-annotated valid and invalid solutions involving long-form reasoning intended to facilitate scalable oversight research (preprint here<\/a>). We’re now finalizing a dataset of textual representations of the research processes followed by high-performing participants in an experiment involving an online research task \u2014 for use in improving LLM capability elicitations \u2014 and writing up the results of our associated experiments. You can sign up<\/a> to be notified when we release future datasets.<\/p>\n

I’m also grateful to have had the opportunity to contribute to Usman et al.’s monumental agenda paper, Foundational Challenges in Assuring Alignment and Safety of Large Language Models<\/a>, and to contribute to some of Anthropic’s Frontier Red Team evaluation\/demo projects as part of collaborations with Hidden Variable Limited.<\/p>\n

See my Google Scholar profile<\/a> for a list of my most cited works, and the bottom of this page<\/a> for recent updates that may not be reflected there.<\/p>\n

Recognition<\/h3>\n