Alec Radford
First author of GPT-2, CLIP and Whisper at OpenAI; independent researcher on 2026 papers
Alec Radford is a machine learning researcher who is first author of the papers behind GPT-2, CLIP and Whisper, whose code and model weights OpenAI released publicly. CLIP learns image representations by matching images with captions, and Whisper is a speech recognition model trained on large-scale weak supervision; both repositories are under the MIT license. Radford also co-authored the GPT-3 paper and the OpenAI scaling-laws paper, whose contribution statement says Radford helped develop the optimized Transformer implementation and the text datasets. On two papers from early 2026, on token-level pretraining data filtering and on generative models of language-model activations, Radford is listed as independent.512346789
Work
CLIPOpen weights
Equal-contribution first author of "Learning Transferable Visual Models From Natural Language Supervision" (2021), which releases code and pre-trained model weights; the repository is under the MIT license.12
WhisperOpen weights
Equal-contribution first author of "Robust Speech Recognition via Large-Scale Weak Supervision" (2022), which releases the models and inference code; the repository is under the MIT license.34
GPT-2 (external site: github.com)Open weights
First author of "Language Models are Unsupervised Multitask Learners" (2019), as cited in the openai/gpt-2 repository, which publishes the code and models under a modified MIT license.5
Scaling Laws for Neural Language Models (external site: arxiv.org)Published research
Co-author of the 2020 paper; its contribution statement credits Radford, with others, with developing the optimized Transformer implementation and the text datasets.7
Other documented roles
Roles the sources document but that are not confirmed as current.
This profile covers public professional work only, from the sources below. It is not an endorsement or a ranking. To correct or remove it, use the correction form.