Microsoft’s new AI can be 4x more accurate and 20% cheaper than human doctors — Mustafa Suleyman calls it “a genuine step toward medical superintelligence”

by

As generative AI continues to grow in popularity and sophistication, many professionals are increasingly worried about job stability beyond just concerns about safety and data protection. It’s becoming clearer with each advancement that such technology may eventually replace certain roles within the workforce.

In simpler terms, Salesforce CEO Marc Benioff has shared that AI is being utilized to automate around half of the company’s workload. You may be aware that the executive mentioned they were contemplating hiring software engineers in 2025, as a result of remarkable increases in productivity due to autonomous AIs.

It appears that the field of medicine might soon find itself under the scrutiny of artificial intelligence, following Microsoft’s announcement that their advanced AI tool, capable of diagnosing diseases with four times greater accuracy than a human physician, has been developed (according to WIRED).

Besides enhancing precision, Microsoft’s novel AI health tool offers another advantage: it’s substantially more affordable than a typical healthcare practitioner’s fee.

As a bystander, I can attest that Microsoft’s AI CEO, Mustafa Suleyman, views this tool as a significant stride towards achieving “medical superintelligence.

The tech titan utilized 304 case studies from the New England Journal of Medicine as a foundation to develop a diagnostic test named Sequential Diagnosis Benchmark. This test was employed to evaluate the performance of their AI tool. By meticulously analyzing each case study, stage by stage, similar to how a real doctor would approach diagnosis, they were able to arrive at a definitive conclusion.

Microsoft’s team created the MAI Diagnostic Orchestrator (MAI-DxO), a system that employs several models such as OpenAI’s GPT, Google’s Gemini, Anthropic’s Claude, Meta’s Llama, and xAI’s Grok. It works by mimicking a real-life situation where multiple specialists (in this case, doctors) collaborate to address queries.

As a researcher, I find it fascinating to report that our newly developed system significantly surpassed human doctors in performance, achieving an outstanding 80%, whereas human doctors only managed a modest 20%. It’s noteworthy that this system also proved economical by cutting down the overall cost for services by 20%, through its judicious choice of cost-effective procedures and tests.

In this study, the MAI-DxO system demonstrated superior performance over human doctors, attaining an accuracy rate of 80%, whereas human doctors achieved only 20%. Additionally, it managed to save 20% on expenses by opting for less costly diagnostic tests and treatments.

The coordination system, consisting of numerous collaborative agents, is the method we believe will lead us towards achieving medical superintelligence by engaging in a debate-like sequence.

Microsoft AI CEO, Mustafa Suleyman

It’s uncertain whether Microsoft intends to make this tool widely available for general use. However, a reliable source close to the company has hinted that they may be incorporating the tool directly into their search engine, Bing, allowing users to quickly and accurately diagnose health problems when needed.

David Sontag, an MIT researcher, finds Microsoft’s latest AI health system intriguing but advises that the results should be approached with some reservation, since the physicians involved in the study were instructed not to employ any extra tools. Additionally, Sontag hints that this setup may not accurately reflect a real-world scenario.

Satya Nadella’s AI dream is taking good shape

Last month, I shared insights on the visions of Microsoft’s CEO Satya Nadella and OpenAI’s CEO Sam Altman regarding an AI-powered future. Notably, it appears that these leaders have recently diverted their attention away from striving for the highly sought-after milestone of Artificial General Intelligence (AGI).

From my perspective, I’ve always been captivated by the concept of a self-replicating spaceship. Yet, in contrast to my fascination, Satya Nadella once dismissed Artificial General Intelligence (AGI) as mere “benchmark hacking.” It appears that he is less interested in theoretical advancements and more focused on leveraging AI for tangible, real-world impact.

In other news, Microsoft’s significant collaboration with OpenAI might face scrutiny due to increasing investor demands for OpenAI to transition into a profit-driven company. Another source indicated that OpenAI, creator of ChatGPT, may accelerate its declaration of Artificial General Intelligence (AGI) to end its association with the tech industry leader prematurely.

Read More

2025-07-01 12:09