Brandeis University
Stanford University
Tom BrownCo-Founder and Chief Compute Officer · AnthropicResearcher↗
Researcher · Founder
Sam McCandlish
AnthropicCo-founder and Chief Architect
McCandlish co-authored scaling laws that help AI labs estimate the gains from more data and computing power before expensive training runs.[1] He also helped develop a way to predict efficient training-batch sizes—the amount of data processed at once—making large-model experiments easier to plan.[2]
Career & education
Current roleAnthropicJan 2021-presentCo-founder and Chief Architect
AnthropicJan 2021-presentPreviously
OpenAI2018-2020Member of Technical Staff
Boston University2017-2018Postdoc
OpenAI2018-2020Boston University2017-2018Work and education history is primarily focused on AI-relevant roles and may not be comprehensive.
Citation Trend
Total citations to scholarly work, including research papers, technical reports, books and theses, as recorded by Google Scholar. Each point is a dated snapshot.
citations157,459As of Nov 1, '26
citations+6,504Since last snapshot, Oct 1, '26
Selected contributions
Language Models are Few-Shot Learners
With Tom Brown, Ben Mann, Nick Ryder, Melanie Subbiah, Jared Kaplan, Prafulla Dhariwal, Arvind Neelakantan and others ·GPT-3 performs tasks from prompts and a few examples without task-specific training.
Scaling Laws for Neural Language Models
With Jared Kaplan, Tom Henighan, Tom Brown, Benjamin Chess, Rewon Child, Scott Gray, Alec Radford and others ·Measured predictable links between language-model performance, model size, data and computation.
Constitutional AI: Harmlessness from AI Feedback
With Yuntao Bai, Saurav Kadavath, Sandipan Kundu, Amanda Askell, Jackson Kernion, Andy Jones, Anna Chen and others ·Trained AI assistants to critique and improve their answers using a written set of principles.
References
- Scaling Laws for Neural Language Modelsarxiv.org
Coauthorship, empirical scaling and compute-allocation planning.
- An Empirical Model of Large-Batch Trainingarxiv.org
Lead coauthorship, useful batch-size prediction and compute/time efficiency tradeoff.
Based on public sources. If accuracy is important, verify details using the links or references.
For corrections, updates, or requests to remove information, contact us.
Reviewed August 7, 2026 · Checked for accuracy and new information.
