{"type":"rich","version":"1.0","provider_name":"Transistor","provider_url":"https://transistor.fm","author_name":"Benevolent Disruptors","title":"Episode 5 | Thomas Wolf, Co-Founder of Hugging Face","html":"<iframe width=\"100%\" height=\"180\" frameborder=\"no\" scrolling=\"no\" seamless src=\"https://share.transistor.fm/e/3105e1a3\"></iframe>","width":"100%","height":180,"duration":2602,"description":"In this episode, Chris Corbishley speaks with Thomas Wolf, co-founder and Chief Science Officer of Hugging Face, the open-source AI platform now powering millions of models and researchers worldwide. Thomas shares the unlikely origins of the company from a teen chatbot to the central hub of the global machine learning community sparked by a research codebase for BERT and GPT-2 that unexpectedly went viral and reshaped the company’s trajectory.He discusses how Hugging Face now incubates internal “mini-startups” across open science, small on-device models, robotics and AI for science, and how this mirrors his work angel-investing in over 100 AI companies. The conversation explores the cultural differences between European and US founders, the importance of mission and openness, the limitations of today’s large language models, and why Thomas believes multiple research paths, not just those pursued by frontier labs, are essential for the field’s long-term progress.Key TakeawaysHow Hugging Face evolved from a teen chatbot into the backbone of the global open-source AI ecosystemWhy an open-sourced research codebase for BERT and GPT-2 became the catalyst for a full company pivotHow Hugging Face incubates internal “startups” such as BigScience, small on-device models, robotics, and AI for scienceWhy Thomas believes mission, culture, and long-term orientation are essential and how they emerged over time rather than being predeterminedHow European founders can overcome self-censorship and think bigger, and why Thomas encourages a more American-style approach to ambitionWhy he sees AI as a dual-use technology, the limitations of current LLMs, and the importance of multiple research paths beyond today’s frontier labsHow regulatory cycles swing between under- and over-correction, and why this matters for innovatorsWhere Thomas sits on the spectrum between benevolent and disruptor—and how that shapes his work today👉 Subscribe to Benevolent Disruptors for more conversations...","thumbnail_url":"https://img.transistorcdn.com/e3_GV3LBH7RGlOeqljrTYvIZI-Lp7_153hlD3SBJPgk/rs:fill:0:0:1/w:400/h:400/q:60/mb:500000/aHR0cHM6Ly9pbWct/dXBsb2FkLXByb2R1/Y3Rpb24udHJhbnNp/c3Rvci5mbS8wMDlm/Y2U3OTljNmQxM2Jk/YzBhNzVhNTQ3OTgz/NzU4Yi5wbmc.webp","thumbnail_width":300,"thumbnail_height":300}