Tom Brown is a machine learning engineer and researcher known as the lead author of the 2020 paper that introduced GPT-3, and as one of the co-founders of Anthropic in 2021.
GPT-3
Before working in AI, Brown had a background as an engineer and entrepreneur, including work related to real-time video and effects technology. He joined OpenAI and became first author of "Language Models are Few-Shot Learners" (2020), the paper that introduced GPT-3, a 175-billion-parameter Autoregressive model trained on a broad web-scale corpus. The paper's central claim, that a sufficiently large language model could perform new tasks from only a handful of examples given in the prompt without any weight updates, established Few-shot learning and In-context learning as central framings for how practitioners would interact with large models going forward, and GPT-3's API access model set the template later followed across the industry.
Anthropic
In 2021, Brown left OpenAI along with several colleagues, including siblings Dario Amodei and Daniela Amodei, to found Anthropic, a company organized around the thesis that frontier AI capabilities and safety research needed to be pursued together rather than treated as separate concerns. As a cofounder, Brown has worked on the engineering and infrastructure needed to train Anthropic's Claude model family, an area related to but distinct from the alignment-focused public messaging most associated with Anthropic's leadership.
Significance
Brown's GPT-3 paper is among the most cited works in the deep learning literature of the 2020s and is frequently credited with popularizing the "scale is what matters" reading of language model progress that also underlies Scaling laws research by colleagues such as Jared Kaplan. The paper's demonstration that a single, general-purpose model could be adapted to new tasks purely through prompting, without any retraining, is often described as the moment prompting itself became a serious technical discipline rather than an afterthought to model training.
His move from OpenAI to Anthropic was part of a cohort departure that also included Chris Olah, and it is frequently cited as one of the clearest examples of researcher migration reshaping the competitive landscape among frontier AI labs. As at OpenAI, Brown has kept a comparatively low public profile at Anthropic relative to its most visible spokespeople, focusing on the engineering work of training and scaling models rather than public communication, a division of labor common among the technical cofounders of frontier labs.