I am a research scientist at Anthropic. My research focuses on sociotechnical alignment of AI systems: how to build AI systems that interact positively with people and the societies we live in. Concretely, this looks like:
Developing scalable evaluations for both present-day and emerging harms (DiscrimEval)
Creating interpretability methods to understand and steer models (Codebook Features)
Designing new ways for humans and AI systems to interact (GATE)
Understanding the limitations of current paradigms; e.g. by training Transformers across a dozen different modalities (DABS and DABS 2.0)
Designing ways of eliciting values from groups of people and training models using those values (Collective Constitutional AI)
If this sounds interesting, we're hiring!
Previously, I completed my PhD in Computer Science at Stanford, where I was advised by Noah Goodman and part of the Stanford AI Lab and Stanford NLP Group.
Selected Publications
Collective Constitutional AI: Aligning a Language Model with Public Input [📝blogpost]
Saffron Huang, Divya Siddarth, Liane Lovitt, Thomas I. Liao, Esin Durmus, Alex Tamkin, Deep GanguliFAccT 2024Press: [New York TImes] [Time Magazine] [Business Insider]Evaluating and Mitigating Discrimination in Language Model Decisions [🐦thread]
Alex Tamkin, Amanda Askell, Liane Lovitt, Esin Durmus, Nicholas Joseph, Shauna Kravec, Karina Nguyen, Jared Kaplan, Deep GanguliArXiv PreprintPress: [VentureBeat] [TechCrunch]Eliciting Human Preferences with Language Models [🐦thread]
Belinda Z. Li*, Alex Tamkin*, Noah D. Goodman, Jacob AndreasArXiv PreprintPress: [VentureBeat]Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet [📝blogpost]
Adly Templeton*, Tom Conerly*, Jonathan Marcus, Jack Lindsey, Trenton Bricken, Brian Chen, Adam Pearce, Craig Citro, Emmanuel Ameisen, Andy Jones, Hoagy Cunningham, Nicholas L Turner, Callum McDougall, Monte MacDiarmid, Alex Tamkin, Esin Durmus, Tristan Hume, Francesco Mosconi, C. Daniel Freeman, Theodore R. Sumers, Edward Rees, Joshua Batson, Adam Jermyn, Shan Carter, Chris Olah, Tom HenighanPress: [New York Times] [WIRED] [TIME]Codebook Features: Sparse and Discrete Interpretability for Neural Networks [🐦thread][📝blogpost]
Alex Tamkin, Mohammad Taufeeque, Noah D. GoodmanICML 2024Towards Measuring the Representation of Subjective Global Opinions in Language Models
Esin Durmus, Karina Nguyen, Thomas I. Liao, Nicholas Schiefer, Amanda Askell, Anton Bakhtin, Carol Chen, Zac Hatfield-Dodds, Danny Hernandez, Nicholas Joseph, Liane Lovitt, Sam McCandlish, Orowa Sikder, Alex Tamkin, Janel Thamkul, Jared Kaplan, Jack Clark, Deep GanguliCOLM 2024Towards Monosemanticity: Decomposing Language Models With Dictionary Learning
Trenton Bricken*, Adly Templeton*, Joshua Batson*, Brian Chen*, Adam Jermyn*, Tom Conerly, Nicholas L Turner, Cem Anil, Carson Denison, Amanda Askell, Robert Lasenby, Yifan Wu, Shauna Kravec, Nicholas Schiefer, Tim Maxwell, Nicholas Joseph, Alex Tamkin, Karina Nguyen, Brayden McLean, Josiah E Burke, Tristan Hume, Shan Carter, Tom Henighan, Chris OlahPreprintFeature Dropout: Revisiting the Role of Augmentations in Contrastive Learning
Alex Tamkin, Margalit Glasgow, Xiluo He, Noah GoodmanNeurIPS 2023Task Ambiguity in Humans and Language Models
Alex Tamkin*, Kunal Handa*, Avash Shrestha, Noah GoodmanICLR 2023Oolong: Investigating What Makes Crosslingual Transfer Hard with Controlled Studies [🐦thread]
Zhengxuan Wu*, Isabel Papadimitriou*, Alex Tamkin*EMNLP 2023DABS 2.0: Improved Datasets and Algorithms for Universal Self-Supervision [🐦thread]
Alex Tamkin, Gaurab Banerjee, Mohamed Owda, Vincent Liu, Shashank Rammoorthy, Noah GoodmanNeurIPS 2022Active Learning Helps Pretrained Models Learn the Intended Task [🐦thread]
Alex Tamkin*, Dat Nguyen*, Salil Deshpande*, Jesse Mu, Noah GoodmanNeurIPS 2022DABS: A Domain-Agnostic Benchmark for Self-Supervised Learning [🌐site] [🐦thread]
Alex Tamkin, Vincent Liu, Rongfei Lu, Daniel Fein, Colin Schultz, Noah GoodmanNeurIPS 2021Press: [Redshift Magazine] [AIM Magazine] [Stanford HAI]C5T5: Controllable Generation of Organic Molecules with Transformers
Daniel Rothchild, Alex Tamkin, Julie Yu, Ujval Misra, Joseph GonzalezArXiv PreprintOn the Opportunities and Risks of Foundation Models
Center for Research on Foundation Models (full list of authors)– Section 4.2: Training and Self-Supervision, Alex Tamkin– Section 4.9: AI Safety and Alignment, Alex Tamkin, Geoff Keeling, Jack Ryan, Sydney von Arx– Coauthor: Sections §2.2: Vision, §3.3: Education, §4.1 Modeling, §5.6: Ethics of ScalePress: [Forbes] [The Economist] [VentureBeat]Viewmaker Networks: Learning Views for Unsupervised Representation Learning [📝blogpost] [🐦thread]
Alex Tamkin, Mike Wu, Noah GoodmanICLR 2021Understanding the Capabilities, Limitations, and Societal Impact of Large Language Models [📝blogpost]
Alex Tamkin*, Miles Brundage*, Jack Clark, Deep GanguliArXiv Preprint Press: [WIRED] [VentureBeat] [Datanami] [Slator]Language Through a Prism: A Spectral Approach for Multiscale Language Representations [🐦thread] [📝blogpost]
Alex Tamkin, Dan Jurafsky, Noah GoodmanNeurIPS 2020Investigating Transferability in Pretrained Language Models [🐦thread]
Alex Tamkin, Trisha Singh, Davide Giovanardi, Noah GoodmanFindings of EMNLP 2020; Presented at CoNLL 2020Distributionally-Aware Exploration for CVaR Bandits.
Alex Tamkin, Ramtin Keramati, Christoph Dann, Emma Brunskill. NeurIPS 2019 Workshop on Safety and Robustness in Decision Making; RLDM 2019Media
Anthropic's YouTube Channel - How we built Artifacts with Claude
The Pragmatic Engineer - How Anthropic built Artifacts
Quanta Magazine - How Quickly Do Large Language Models Learn Unexpected Skills?
VentureBeat - Anthropic leads charge against AI bias and discrimination with new research
TechCrunch - Anthropic’s latest tactic to stop racist AI: Asking it ‘really really really really’ nicely
VentureBeat - How can AI better understand humans? Simple: by asking us questions
WIRED Magazine - Chatbots Got Big—and Their Ethical Red Flags Got Bigger
Abrupt Future Podcast - Alex Tamkin on ChatGPT and Beyond: Navigating the New Era of Generative AI
AI Artwork in PC Magazine (twitter thread: DALL-E Meets WALL-E: an Art History)
The Gradient Podcast - Alex Tamkin on Self-Supervised Learning and Large Language Models
Press: [Communications of the ACM]Personal
Other topics I think a lot about:
Societal impacts of technology, especially machine learning and large language models
Mentoring, teaching and fostering a healthy and inclusive research culture
Scientific communication and breaking down walls between fields
I also like making art, especially ceramics and photography!