When people talk about Anthropic, CEO Dario Amodei is usually the name that gets attention. But another co-founder of the AI company has an equally interesting career journey. Chris Olah dropped out of college at 18, received $100,000 (roughly Rs 93 lakh) from billionaire investor Peter Thiel and went on to work at Google and OpenAI before helping build Anthropic.Today, Olah serves as Anthropic's co-founder and interpretability research lead. Unlike people who focus on making AI models more powerful, much of his work has centred on understanding how these models function internally and why they produce particular responses.How did he get jobs at Google, OpenAI without a college degree?His journey began in Toronto, Canada, where he enrolled at the University of Toronto to study mathematics. However, he left the university after roughly a year. In 2012, he received a $100,000 (roughly Rs 93 lakh) Thiel Fellowship, a program founded by PayPal co-founder Peter Thiel that supports young people who want to pursue their interests instead of following the traditional college route.Olah had already developed an interest in machine learning and explaining complicated technical concepts. He began publishing detailed explainers on his personal blog, using visualisations to make the workings of neural networks easier to understand. His work gradually attracted attention from researchers in the machine learning community. His research work and growing reputation in the machine learning community helped him get opportunities at some of the biggest names in technology, despite not completing a university degree. Olah's first major opportunity came at Google Brain, where he joined as an intern in 2015. He subsequently became a research associate and then a research scientist, spending more than three years with the team.At Google, Olah worked on ways to visualise neural networks and understand how they process information. This area of research, known as mechanistic interpretability, attempts to explain what happens inside AI models instead of treating them as black boxes. In September 2018, he joined OpenAI as a member of its technical staff and eventually led its interpretability team. During his time there, he worked on two notable research projects.One was the Circuits project, which explored how groups of neurons work together to perform specific tasks. The other involved the discovery of multimodal neurons in OpenAI's CLIP model, which connects images and text. Researchers found that certain neurons responded to the same concept across different formats. For instance, a neuron associated with Spider-Man could activate when the model encountered a photograph, a drawing or text about the character.The research offered a better understanding of how neural networks recognise and process information.Why did he leave OpenAI to co-found Anthropic?Olah left OpenAI in December 2020 along with a group of colleagues who had concerns about AI safety. The group included Dario Amodei, who later became Anthropic's CEO. In March 2021, Olah joined the group in co-founding Anthropic. He took up the role of co-founder and interpretability research lead, continuing the work he had pursued at Google and OpenAI.At Anthropic, his research focuses on understanding the internal workings of AI models, including identifying which groups of artificial neurons activate for particular concepts and how they influence a model's responses. The broader goal is to help researchers understand these systems better and assess their behaviour and safety. His work has also received recognition outside the research community. In 2024, Time included Olah in its TIME100 AI list of influential people in the artificial intelligence industry.- Ends
He dropped out of college, worked at Google and OpenAI, then co-founded Anthropic
Full Article
Original Source
Read the full article at Indiatoday →KhanList aggregates and links to publicly available news content. We do not host full articles from third-party sources. Always verify important information with original sources.