Microsoft harnesses AI to accelerate breakthroughs in scientific discovery

Can artificial intelligence accelerate parts of the scientific process? Microsoft certainly seems to believe so.

At its Build 2025 conference on Monday, Microsoft introduced Microsoft Discovery, a platform leveraging agentic AI to "transform the [scientific] discovery process," according to a press release shared with TechCrunch. Microsoft describes the platform as "extensible" and capable of managing certain scientific workloads "end-to-end."

"Microsoft Discovery is an enterprise agentic platform designed to accelerate research and discovery by reshaping the entire process with agentic AI—from scientific knowledge reasoning and hypothesis formulation to candidate generation, simulation, and analysis," Microsoft explains in the release. "The platform allows scientists and researchers to collaborate with a team of specialized AI agents, helping drive scientific outcomes with greater speed, scale, and accuracy through the latest advances in AI and supercomputing."

Microsoft is among many AI labs optimistic about AI's potential in science. Earlier this year, Google introduced an "AI co-scientist," which the company claimed could assist researchers in developing hypotheses and research plans. Anthropic, its main competitor OpenAI, as well as organizations like FutureHouse and Lila Sciences, have argued that AI tools could significantly speed up scientific discovery, particularly in medicine.

However, many researchers currently view AI as less effective in guiding the scientific process, largely because of its unreliability.

One of the main challenges in creating an "AI scientist" is accounting for countless unpredictable variables. AI may prove useful in areas requiring broad exploration, such as narrowing down extensive lists of possibilities, but it remains uncertain whether it can deliver the kind of innovative problem-solving that leads to genuine breakthroughs.

So far, results from AI systems designed for scientific applications have been largely disappointing.

In 2023, Google reported that approximately 40 new materials had been synthesized with the help of its AI tool, GNoME. However, external analysis revealed that none of those materials were actually new. Meanwhile, several companies using AI for drug discovery, including Exscientia and BenevolentAI, have experienced notable clinical trial failures.

Join Us at TechCrunch Sessions: AI

Secure your spot at our premier AI industry event, featuring speakers from OpenAI, Anthropic, and Cohere. For a limited time, tickets are only $292 for a full day of expert talks, workshops, and high-impact networking.

Exhibit at TechCrunch Sessions: AI

Reserve your place at TC Sessions: AI and showcase your innovations to 1,200+ decision-makers—without the high cost. Available until May 9 or while tables remain.

Microsoft undoubtedly hopes its initiative will achieve greater success than previous efforts.