Keep pulling the thread on Sam Rodriques.
Over time, AI model performance is expected to improve to the point where they will possess deep knowledge across many domains, potentially surpassing human capabilities in generating scientific hypotheses.
Future House's AI platform identified a novel hypothesis for treating age-related macular degeneration (AMD) by connecting previously disparate findings in the scientific literature.
The AI platform from Future House discovered a potential treatment for AMD by identifying a link between Rho-kinase (ROCK) inhibitors, known in immunology to affect phagocytosis, and the pathophysiology of AMD, a connection not previously made in ophthalmology literature.
An existing FDA-approved Rho-kinase (ROCK) inhibitor was identified by Future House's AI as a potential direct treatment for age-related macular degeneration (AMD).
In a wet lab cell culture assay for age-related macular degeneration (AMD), the Rho-kinase (ROCK) inhibitors identified by Future House's AI worked very well, performing even better than the positive controls.
Future House is a San Francisco-based research lab focused on building an AI scientist to automate scientific discovery and research in biology and other complex sciences.
Future House's AI platform is built on the concept of "true agents," which are language models given access to scientific tools that can make decisions and take actions within an environment.
Future House has developed several specialized AI agents, including Crow for general literature search, Falcon for deep literature synthesis, Finch for data analysis, Owl for precedent search, and Phoenix for chemistry experiment planning.
The AI agents developed by Future House are designed as a multi-agent system where they can delegate tasks and communicate with each other to solve complex scientific problems.
Currently, the best human experts in any given field still possess better judgment than any existing AI agent or language model.
AI agents' primary advantage over human experts today is their ability to integrate information across a vast number of different fields, whereas humans tend to have deep but narrow expertise.
Even with a hypothetical perfect cure for aging discovered today, it would take at least 5 to 10 years to confirm its efficacy due to the fundamental timescale of human aging.