Futurehouse, a nonprofit organization supported by Eric Schmidt, aimed at building “AI scientists” within the next decade, has released a new tool that claims to be able to support “data-driven discovery” in biology. The new tool comes just a week after Futurehouse launched its API and platform.
A tool called Finch incorporates biological data (mainly the form of research papers) and prompts (“Please tell me about the molecular drivers for cancer matese?”) and runs the code before generating numbers and examining the results. In a series of posts on X, we compared Futurehouse co-founder and CEO Sam Rodriques with a “first-year alumni.”
“It’s a superpower that can (do all of this) in minutes,” writes Rodriquez. “(Finch) will actually find some really cool things (…) for our own projects. We found it to be pretty great.”
Futurehouse’s proposal is that, like many startups and tech giants, Finch and other AI tools will one day automate steps in the science process.
In an essay earlier this year, Openai CEO Sam Altman said that “urgent” AI tools “can “rapidly accelerate scientific discovery and innovation.” Similarly, Anthropic CEO, who launched the AI for Science program this week, boldly predicts that AI will help develop treatments for most cancers.
However, the evidence is lacking. Many researchers do not believe that AI today is particularly useful in guiding scientific processes. Needless to say, Futurehouse has yet to achieve any scientific breakthrough or make novel discoveries with its AI tools.
Biology, particularly on the drug discovery side, is an attractive target for AI companies. Priority survey estimates that the market was worth $65.888 billion in 2024, potentially reaching $16003.1 billion by 2034.
TechCrunch Events
Berkeley, California
|
June 5th
Book now
There have been some successes, but AI does not offer an immediate magical solution in the lab. Several companies employing AI for drug discovery, including Exscientia and Benevolentai, have suffered from the failure of well-known clinical trials in recent years. Meanwhile, the accuracy of major AI systems for drug discovery, such as Google Deepmind’s Alphafold 3, tends to vary widely.
Similarly, Finch is making a “silly mistake,” Rodriquez said. That is why Future House is recruiting bioinformatic biologists and computational biologists to assess and train their accuracy and reliability during a closed beta.
Interested people can sign up here.