FutureHouse introduces artificial intelligence tools that they claim can accelerate scientific research

This Eric Schmidt-backed nonprofit organization aims to create “scientist AI” over the next decade and recently released its first significant product: a platform and API with AI-based tools designed to support scientific endeavors.
An AI platform and API with AI tools designed to support scientific endeavors.
Many startups are actively developing tools to apply AI to science, many of which are backed by significant venture capital funding. Technology giants are also showing optimism about the use of AI in science. For example, earlier this year, Google unveiled an “AI scientist” that the company says can help scientists formulate hypotheses and design experimental research plans.
Some startups are actively developing tools to apply AI to science, many with significant venture funding.
The CEOs of OpenAI and Anthropic argue that AI tools can greatly accelerate scientific discovery, especially in the medical field. However, many researchers still don’t see AI as particularly useful for managing the scientific process because of its unreliability.
OpenAI and Anthropic CEOs argue that AI tools can greatly accelerate scientific discovery, especially in the medical field.

FutureHouse announced four new AI tools on Thursday: the Crow, Falcon, Owl, and Phoenix. Crow is capable of searching scientific literature and answering questions about it; Falcon conducts deeper literature searches, including specialized scientific databases; Owl investigates previous work in a given field; and Phoenix uses tools to plan chemistry experiments.
“Today we are launching the first publicly available AI Scientist through the FutureHouse platform. Our AI Scientist agents can perform a wide range of scientific tasks better than humans. By combining their capabilities, we’ve already begun to rapidly break new ground in biology,” said Sam Rodriguez.
FutureHouse emphasizes that their tools have access to an extensive corpus of high-quality open access articles and specialized scientific tools. “They also provide transparent reasoning and use a multi-step process to analyze each source in greater depth. By combining these AIs at scale, scientists can dramatically accelerate the pace of scientific discovery,” the company’s blog notes.
Scientists can significantly accelerate the pace of scientific discovery,” the company said.
It’s worth noting, however, that FutureHouse has yet to achieve significant scientific breakthroughs or new discoveries with its AI tools.
One of the challenges of developing “scientist AI” is the many confounding factors. AI can be useful in areas with a wide range of research, such as selecting from a variety of possibilities. However, it is unclear whether it is capable of finding out-of-the-box solutions to problems that lead to real breakthroughs.
It is also unclear whether it is capable of finding out-of-the-box solutions to problems that lead to real breakthroughs.
The results of AI systems for science are not yet impressive. In 2023, Google reported synthesizing about 40 new materials using its GNoME AI, but external analysis has not revealed any truly new materials.
The technical shortcomings and risks of using AI – such as its propensity for error – have scientists concerned about its use in serious research. Even carefully planned studies can be marred by incorrect actions by AI, which has difficulty performing highly accurate tasks.
An AI that is not able to perform highly accurate tasks can be used for research.
FutureHouse recognizes the potential for bugs in its AI tools – especially Phoenix. “We’re launching this now in the spirit of rapid iteration,” the company notes on its blog. “Please provide feedback as you use it.”
The story FutureHouse unveils artificial intelligence tools they claim can accelerate scientific research was first published on ITZine.ru.