Alec Radford is an American machine learning researcher who was one of the most influential figures behind OpenAI’s generative models. Before joining OpenAI in 2016, he co-founded the startup Indico and, in November 2015, was first author of “Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks” (with Luke Metz and Soumith Chintala). That paper introduced DCGAN, a stable convolutional recipe for generative adversarial networks that became one of the most widely cited templates in image generation.
At OpenAI, Radford was the lead author on the original GPT paper in 2018 (“Improving Language Understanding by Generative Pre-Training”) and on GPT-2 in 2019 (“Language Models are Unsupervised Multitask Learners”), establishing the generative-pretraining recipe that underpins ChatGPT. He was also first author on CLIP, which connected images and text through contrastive learning, and on Whisper, OpenAI’s robust multilingual speech-recognition system. His name appears on the Scaling Laws and Jukebox papers as well, making him a thread running through much of OpenAI’s early breakthrough research.
Radford left OpenAI in late 2024 and subsequently became associated with Thinking Machines Lab, the startup founded by former OpenAI CTO Mira Murati.
Why business readers should care: A handful of researchers shaped the architectures now embedded in everyday products, and Radford’s fingerprints are on an unusually large share of them. Understanding who built the foundational models helps explain why talent moves between labs command so much attention.