Source: Fortune
Summary
Sara Hooker, a former AI researcher at Google and Cohere, has raised $50 million in seed funding for her new startup, Adaption Labs. The company aims to create AI systems that use less computing power and cost less to run than current leading models. Hooker and co-founder Sudip Roy are targeting models that can learn continuously without expensive retraining or fine-tuning. Adaption Labs is exploring new AI architectures, including gradient-free learning, to achieve this goal.
Our Reading
The strategy enters a familiar phase.
Adaption Labs is betting against the prevailing AI industry wisdom that bigger models are better. Hooker argues that the approach is seeing diminishing returns. The company is exploring new AI architectures, including gradient-free learning, to create models that can learn continuously. The funding round is led by Emergence Capital Partners, with participation from several other venture capital firms. Hooker’s goal is to “eliminate prompt engineering” and make AI models more adaptable and efficient.
The numbers tell a story of a market looking for alternatives to the “scale is all you need” approach. As Hooker said, “This is probably the most important problem that I’ve worked on.” The question is whether Adaption Labs can deliver on its promise of more efficient and adaptable AI models.
Author: Evan Null
