FutureHouse, based in San Francisco, is a non-profit research laboratory established to build “AI Scientists” to provide solutions to humanity’s biggest problems. In their 2023 published goals, they aim to accelerate next-generation discoveries by automating the “scientific method,” primarily in biology (https://www.futurehouse.org/)
Today, the volume of scientific production has far exceeded human capacity. There are over 38 million scientific articles in the PubMed database alone as of 2024. Thousands more are added every day. It is becoming increasingly difficult for human researchers to generate meaningful hypotheses, evaluate research, and achieve groundbreaking discoveries in this vast pool of information. This is where FutureHouse comes into play:
FutureHouse adopts a 4-layered approach to robotize scientific discovery:
- AI Tools (AlphaFold-like models, laboratory experiments).
- AI Scientific Assistants (text-based literature research, hypothesis generation).
- AI Scientists (experiment design, hypothesis generation, data analysis, and conclusion writing).
- Human — the stage where scientific “inquiry” processes can be carried out without the need for humans in the future.
This structure makes the scientific process both in-depth and scalable, allowing researchers to make faster discoveries.
A newly published Nature article shared that the model called ether0 developed by FutureHouse has even exceeded human levels in chemistry problems (https://www.nature.com/articles/d41586-025-01753-1). Ether0 answers questions step by step with a logical chain by adopting an approach called “reasoning model”. This technology, unlike traditional LLMs, claims to provide scientific credibility by explaining the “decision process”.
FutureHouse is a completely non-profit “moonshot” laboratory supported by donors such as Eric Schmidt. Its aim is not to change the course of big pharmaceutical companies; It is stated as creating a transparent and disciplined scientific production process dedicated to serving humanity (https://www.nature.com/articles/d41586-025-01753-1).
According to TechCrunch, FutureHouse’s new AI agent named Finch can “work like a grad student” and analyze biological data directly (https://techcrunch.com/2025/05/06/futurehouse-previews-an-ai-tool-for-data-driven-biology-discovery/). Such tools have the potential to expand both the speed and capacity of the laboratory infinitely.
Originally designed as “human intelligence augmentation,” these systems aim to eventually produce “AI Scientists” who can replace humans.
With this project by FutureHouse, a “scientific automation machine” that will scale discoveries from 0 to 1 in the scientific world may emerge. This approach is also inspiring for the academic solidarity platform that advocates that science should serve democracy and humanity. No matter how much technology develops, the mission of being “science for humanity” can be achieved when supported by structures that include human values, ethics and the spirit of autonomous science.
