Alec Radford
Alec Radford is a researcher whose shared work helped develop generative pretraining across text, images, and speech. His early papers studied image representations learned without labels and a language model with a sentiment-related feature. He was a lead author of the early GPT work and shared first-author credit on GPT-2 with Jeffrey Wu. He later co-authored image pixel prediction research, led the CLIP author list, and contributed to Whisper. These projects tested ways to learn useful features or tasks from broad collections of data. He was an advisor at Thinking Machines Lab in 2025.
- Works
- 35
Publication numbers from OpenAlex, as of 14 October 2026. Citations and h-index are hidden because this OpenAlex record may be incomplete or mix in other authors.
Timeline
-
2015
Learning from images
Radford co-authored the DCGAN paper in 2015.
-
June 2018
Generative pretraining
Radford helped introduce generative pretraining in June 2018.
-
February 2019
GPT-2
Radford and Wu shared first-author credit on GPT-2 in 2019.
-
2021
Connecting pictures and words
Radford was first author of the CLIP paper in 2021.
-
2022
Whisper
Radford co-authored Whisper in 2022.
-
2025
An advisory role
Radford was an advisor to Thinking Machines Lab in 2025.
Sources and credits
Photo: Alec Radford; public profile image on X. License: Not stated by source. Source: x.com. Cropped to a square and resized. Ask us to remove or replace this photo.
- Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks
- Learning to Generate Reviews and Discovering Sentiment
- Improving language understanding with unsupervised learning
- Language Models are Unsupervised Multitask Learners
- Generative Pretraining from Pixels
- Learning Transferable Visual Models From Natural Language Supervision
- Robust Speech Recognition via Large-Scale Weak Supervision
- Thinking Machines Lab: 2025 indexed team listing