Microsoft has introduced a bold new initiative to transform scientific research using artificial intelligence. At its Build 2025 conference on Monday, the tech giant unveiled Microsoft Discovery, a platform designed to integrate AI directly into the heart of scientific inquiry.
The enterprise-grade platform allows AI to analyze existing scientific knowledge, propose hypotheses, and even simulate experiments. Microsoft says the system will enable scientists and researchers to collaborate with AI agents to achieve faster, more accurate results. The company believes this approach could overcome traditional bottlenecks in research workflows.
Microsoft is not alone in this pursuit. Earlier this year, Google unveiled AI-powered co-researchers, claiming they can assist with hypothesis development and experiment planning. Other firms—including Anthropic, OpenAI, FutureHouse, and Lila Sciences—have emphasized AI’s potential to accelerate discovery across multiple scientific disciplines.
Despite these claims, many researchers remain cautious. While AI can help narrow down research directions or identify patterns in data, its ability to drive fundamental innovation remains unproven. According to TechCrunch, Google’s much-touted synthesis of 40 new materials through AI was later challenged by analysts, who found none of the materials to be entirely novel. AI-driven drug discovery efforts by Exscientia and BenevolentAI have also encountered failures in clinical trials.
Microsoft insists that AI can reshape the paradigm of scientific research, but whether it can consistently deliver breakthrough innovations is still unclear. The scientific community is watching closely to see if Discovery can live up to its promise—or if it will become another high-profile example of AI’s limits in real-world science.