News Summary

Sara Hooker, former VP of AI Research at Cohere, has launched a new startup called Adaption Labs, aiming to challenge the prevailing trend of scaling large language models (LLMs) in AI development. Alongside Sudip Roy, a fellow veteran from Cohere and Google, Hooker is focusing on building AI systems that can learn and adapt from real-world experiences efficiently. This approach contrasts with traditional methods that rely heavily on scaling up computational resources, a practice some researchers believe is reaching its limits. Adaption Labs seeks to demonstrate that learning from experience could be more cost-effective, potentially changing who controls and benefits from AI technologies. The startup has reportedly closed a seed funding round between $20 million and $40 million and plans to hire globally, emphasizing access to underrepresented regions. If successful, Adaption Labs’ strategy could significantly impact the AI industry’s current trajectory toward scaling LLMs in pursuit of general intelligence.