Meta tags:
author= ;
description=Postdoctoral Research Scientist at FAIR, part of Meta AI. Interested in machine learning fairness, reproducibility for improved efficiency, and metascience. ;

Headings (most frequently used words):

samuel, bell, selected, publications,

Text of the page (most frequently used words):
and (10), the (10), #samuel (8), bell (8), pdf (5), html (5), research (5), machine (3), learning (3), for (3), cambridge (3), 2022 (2), processing (2), neil (2), lawrence (2), 2023 (2), levent (2), sagun (2), bias (2), 2025 (2), more (2), data (2), reasoning (2), llms (2), you (2), interested (2), about (2), between (2), credit (2), scientist (2), with (2), fair (2), completed (2), university (2), computer (2), work (2), copyright, 2026, advances, neural, information, systems, neurips, onno, kampman, jesse, dodge, modeling, multiverse, acm, conference, fairness, accountability, transparency, facct, simplicity, leads, amplified, performance, disparities, 2024, diane, bouchacourt, reassessing, validity, spurious, correlations, benchmarks, north, american, chapter, association, computational, linguistics, naacl, megan, richards, eduardo, sanchez, mariano, coria, meglioli, authors, role, speech, reducing, toxicity, detection, review, equal, contribution, polina, kirichenko, mark, ibrahim, kamalika, chaudhuri, abstentionbench, fail, unanswerable, questions, selected, publications, what, discussing, collaborations, opportunities, please, drop, line, email, samueljamesbell, gmail, dot, com, love, hear, from, simulated, financial, crises, market, risk, built, new, retail, banks, developed, next, generation, scores, kudos, thought, goldman, sachs, previously, was, meta, fundamental, lab, which, joined, after, postdoc, paris, phd, part, supervised, have, also, studied, obtained, master, natural, language, bachelor, science, manchester, laboratory, alan, turing, institute, prof, group, broadly, focused, evaluating, improving, advanced, capabilities, particularly, post, training, methods, eliciting, complex, behaviours, such, uncertainty, adapting, subjective, user, preferences, often, emphasis, scalability, through, use, synthetic, consistent, theme, throughout, llm, reliability, including, understanding, model, behaviour, novel, out, distribution, scenarios, sam, member, technical, staff, where, mental, health, split, time, london, slingshot,

Text of the page (random words):
samuel j bell samuel j bell research scientist i m sam a member of technical staff at slingshot ai where i work on ai for mental health i split my time between london and cambridge uk my research is broadly focused on evaluating and improving advanced capabilities of llms i m particularly interested in post training methods for eliciting complex behaviours such as reasoning about uncertainty or adapting to subjective user preferences often with an emphasis on scalability through the use of synthetic data a consistent theme throughout my work is llm reliability including understanding model behaviour in novel or out of distribution scenarios previously i was a research scientist at fair meta s fundamental ai research lab which i joined in 2023 after a postdoc with fair paris in 2022 i completed my phd in machine learning as part of the ml cl group at the university of cambridge supervised by prof neil lawrence i have also studied at the alan turing institute obtained a master s in natural language processing at the cambridge computer laboratory and completed my bachelor s in computer science at the university of manchester in between i ve simulated financial crises in market risk at goldman sachs built new retail banks at thought machine and developed next generation credit scores at credit kudos if you re interested in learning more about what i do or discussing research collaborations or opportunities please do drop me a line my email is samueljamesbell at gmail dot com i d love to hear from you selected publications abstentionbench reasoning llms fail on unanswerable questions polina kirichenko mark ibrahim kamalika chaudhuri and samuel j bell in review equal contribution 2025 html pdf on the role of speech data in reducing toxicity detection bias samuel j bell megan richards eduardo sanchez mariano coria meglioli and 4 more authors in north american chapter of the association for computational linguistics naacl 2025 html pdf reassessing the validity of spurious correlations benchmarks samuel j bell diane bouchacourt and levent sagun 2024 html pdf simplicity bias leads to amplified performance disparities samuel j bell and levent sagun in acm conference on fairness accountability and transparency facct 2023 html pdf modeling the machine learning multiverse samuel j bell onno kampman jesse dodge and neil lawrence in advances in neural information processing systems neurips 2022 html pdf copyright 2026 samuel j bell