Andrej Karpathy, a prominent figure in the field of artificial intelligence, has officially joined Anthropic, a rising player in AI technology. Karpathy announced his new role on X, expressing enthusiasm for working on large language models (LLMs) and the exciting research and development opportunities that lie ahead.
Karpathy commenced his position this week, where he will lead a team focused on pre-training, headed by Nick Joseph. This phase is crucial for developing LLMs like Claude, as it involves extensive training runs that establish a model’s foundational knowledge and abilities. According to Anthropic, this process is not only vital but also resource-intensive, requiring significant computational power and investment.
An Anthropic spokesperson confirmed that Karpathy’s team will concentrate on harnessing Claude to accelerate research in pre-training, indicating a strategic move by Anthropic. This decision underscores the company’s commitment to AI-assisted research as a key differentiator against competitors like OpenAI and Google.
Karpathy’s background is notable. He was instrumental at OpenAI, where his work focused on deep learning and computer vision. After leaving OpenAI in 2017, he took on a leading role in Tesla’s Full Self-Driving and Autopilot initiatives until his departure in 2022. He then returned to OpenAI for a year before establishing Eureka Labs, an initiative aimed at applying AI in educational settings.
Despite his transition to Anthropic, it remains uncertain how Karpathy’s role at Eureka Labs will evolve, as he has not provided updates on the startup since its launch. Nevertheless, he continues to be passionate about education and plans to revisit this work when the opportunity arises. He has also created educational content, including an online course on neural networks, and maintains a YouTube channel dedicated to lectures on AI and LLMs.
Adding to the momentum at Anthropic is Chris Rohlf, who has joined the company’s frontier red team, aimed at testing AI models against potential threats. Rohlf brings over two decades of experience in cybersecurity, having had roles at Yahoo and Meta, and has previously worked on projects related to security and artificial intelligence.
In a statement on X, Rohlf remarked on the potential for AI to enhance cybersecurity and expressed his excitement about joining Anthropic at such a significant time. The combined expertise of Karpathy and Rohlf signals a robust commitment by Anthropic to push the boundaries of AI technology in both research and security domains.
