Finch, FutureHouse’s AI that Interprets Biology Through Data | Google introduction to generative ai | Generative ai in banking mckinsey | Generative ai use cases in healthcare | Turtles AI

Finch, FutureHouse’s AI that Interprets Biology Through Data
Eric Schmidt-funded new tool analyzes scientific publications to support biological research, but remains under observation for the reliability of the results and actual usefulness in the scientific process
Isabella V7 May 2025

 

FutureHouse, a nonprofit backed by Eric Schmidt, has unveiled Finch, an advanced AI tool designed to support biological research through the automated analysis of scientific data.

Key points:

  • Finch processes scientific papers and answers complex questions by generating analyses and visualizations.
  • FutureHouse aims to develop an “AI Scientist” capable of formulating hypotheses and conducting experiments autonomously.
  • Despite progress, AI in drug discovery still presents significant challenges in terms of reliability.
  • FutureHouse collaborates with experts to improve the accuracy and effectiveness of its AI tools.


Finch is an AI agent developed by FutureHouse, a San Francisco-based nonprofit founded in 2023, with the goal of automating scientific research, particularly in the field of biology. Funded primarily by Eric Schmidt, former CEO of Google, the organization is led by Sam Rodriques, a theoretical physicist with experience in advanced technologies such as spatial transcriptomics and gene therapy.

Finch analyzes biological data, mostly scientific papers, and answers complex questions by generating code, visualizations, and interpretations of the results. Rodriques describes him as a “first-year graduate student,” capable of performing basic science tasks but still susceptible to errors.

FutureHouse has also developed other tools, such as PaperQA2, which achieved state-of-the-art performance on the RAG-Arena science benchmark, and created WikiCrow, a project to generate Wikipedia articles for every protein-coding gene in the human genome.

Despite progress, AI in drug discovery still faces significant challenges. Companies like Exscientia and BenevolentAI have experienced clinical failures, and tools like DeepMind’s AlphaFold 3 show variations in accuracy. To address these challenges, FutureHouse is working with bioinformaticians and computational biologists to improve the reliability of its tools.

FutureHouse’s long-term goal is to develop an “AI Scientist” capable of formulating hypotheses, conducting experiments, analyzing data, and writing scientific papers autonomously. This approach aims to overcome the bottleneck represented by human effort in scientific research, accelerating discovery and innovation.

In a context in which biology represents a field full of mysteries to be explored, tools like Finch could become fundamental to address future scientific challenges.