Andrej Karpathy joins Anthropic's pretraining team; Says, "I think the next few years at the frontier of LLMs will be especially formative."

Andrej Karpathy, an influential figure in AI, has joined Anthropic's pretraining team, starting this week. Previously a co-founder at OpenAI and director at Tesla, he is expected to lead efforts to enhance Claude models' capabilities.

“I think the next few years at the frontier of LLMs will be especially formative,” he stated, showcasing his optimism for the evolving landscape of language models.

Karpathy's role focuses on pre-training, a critical phase that includes the large-scale training necessary for enabling AI models to acquire extensive knowledge. An Anthropic spokesperson emphasized that he will build a team aimed at leveraging Claude to accelerate research in this area, positioning the company competitively against rivals like OpenAI and Google.

His experience, especially in areas like computer vision, was lauded by Elon Musk, who described Karpathy as “arguably the #2 guy in the world in computer vision.” As AI continues to develop, Karpathy's focus on utilizing AI-assisted research is a strategic shift towards innovation rather than pure computational power.

“Excited to welcome Andrej to the Pretraining team!” said team lead Nick Joseph. “He'll be building a team focused on using Claude to accelerate pretraining research itself. I can't think of anyone better suited to do it.”

Andrej Karpathy, co-founder of OpenAI and former Tesla AI executive, has joined Anthropic's pretraining team. Karpathy expressed enthusiasm about his new role, stating, "I think the next few years at the frontier of LLMs will be especially formative," as he begins work on the Claude models.