FutureHouse, a nonprofit supported by Eric Schmidt, aims to build “AI Scientists” over the next decade, launches its first major product: the platform and API, with its AI-driven tools designed to support scientific work.

Many startups are competing to develop AI research tools for the scientific field, with some of them having a lot of VC funding behind them. Technology giants seem to be optimistic about scientific AI. Earlier this year, Google launched the “AI Co-scientist,” which the company said could help scientists develop hypotheses and experimental research plans.

AI company Openai and the CEO of Humans asserted that AI tools can greatly accelerate scientific discoveries, especially in medicine. However, many researchers today do not consider AI particularly useful in guiding scientific processes, largely due to its unreliability.

Future House released four AI tools on Thursday: Crow, Falcon, Owl and Phoenix. Crows can search scientific literature and answer questions about it; falcons can conduct more in-depth literature searches, including scientific databases; owls look for previous work in a given subject area; Phoenix uses tools to help plan chemistry experiments.

Today, we launch the first publicly available AI scientist through the housing platform of the future.

Our AI scientist agents can perform a variety of scientific tasks better than humans. By linking them together, we have begun to discover new biology very quickly. and… pic.twitter.com/wmmzogzpi

-Sam Rodriques (@sgrodriques) May 1, 2025

“Unlike other (AIS), Future House has access to a large number of high-quality open papers and professional scientific tools,” Future House writes in a blog post. “They (also) have transparent reasoning and use a multi-stage process to consider each source to link these (AIs) more deeply (…) together, and large-scale scientists can greatly accelerate the pace of scientific discovery.”

But it means that future housing has not yet achieved scientific breakthroughs or made novel discoveries using its AI tools.

Part of the challenge in developing “AI scientists” is that there are many confounding factors expected. AI may come in handy in areas that require extensive exploration, such as narrowing down a large list of possibilities. However, it is not clear whether AI can achieve out-of-the-box questions that lead to a truly breakthrough breakthrough.

So far, the results of AI systems designed for science have been mostly incredible. In 2023, Google said that with the help of one of its AISs, about 40 new materials have been synthesized, called Gnome. However, external analysis found that in fact, no material was new.

The technological shortcomings and risks of AI, such as the tendency of hallucinations, also make scientists cautious about their recognition of serious work. Even well-designed research can end up being tainted by misbehaving AI, which struggles to perform high-precision work.

Indeed, future homes acknowledge that their AI tools (especially Phoenix) may make mistakes.

“We are publishing (now) in the spirit of rapid iteration,” the company wrote in its blog post. “Please provide feedback as you use it.”