Microsoft's new AI tool a medical genius? Tech giant claims it is 4x more accurate than real doctors
medical AI
tool that performs better than human doctors at complex health diagnoses, creating a 'path to medical superintelligence'. The Microsoft AI team shared research that demonstrated how AI can sequentially investigate and solve medicine's most complex diagnostic challenges—cases that expert physicians struggle to answer.
Tech company's AI unit, led by the British tech pioneer
Mustafa Suleyman
, has developed a system that imitates a panel of expert physicians tackling 'diagnostically complex and intellectually demanding' cases.
Microsoft AI Diagnostic Orchestrator
(MAI-DxO) correctly diagnosed up to 85% of NEJM case proceedings, a rate more than four times higher than a group of experienced physicians. MAI-DxO also gets to the correct diagnosis more cost-effectively than physicians, the company said in a blog post.
ALSO READ:
Microsoft layoffs: Tech giant's sales head Judson Althoff asked to go on two-month leave. Here's why
Microsoft says AI system better than doctors
The Microsoft AI Diagnostic Orchestrator', or MAI-DxO for short, the AI-powered tool is developed by the company's AI health unit, which was founded last year by Mustafa Suleyman. The tech giant said when paired with OpenAI's advanced o3 AI model, its approach 'solved' more than eight of 10 case studies specially chosen for the diagnostic challenge. When those case studies were tried on practising physicians – who had no access to colleagues, textbooks or chatbots – the accuracy rate was two out of 10. Microsoft said it was also a cheaper option than using human doctors because it was more efficient at ordering tests.
When benchmarked against real-world case records, the new medical AI tool 'correctly diagnoses up to 85% of NEJM case proceedings, a rate more than four times higher than a group of experienced physicians' while being more cost-effective.What's impressive is that these cases are from the New England Journal of Medicine and are very complex and require multiple specialists and tests before doctors can reach any conclusion.
Live Events
According to The Wired, the Microsoft team used 304 case studies sourced from the New England Journal of Medicine to devise a test called the Sequential Diagnosis Benchmark. A language model broke down each case into a step-by-step process that a doctor would perform in order to reach a diagnosis.
ALSO READ:
Melania should be on first boat: Deportation calls for US' First Lady gains traction amid Trump's immigration crackdown
Microsoft new AI tool diagnosed 85% cases
For this, the company used different large language models from OpenAI, Meta, Anthropic, Google, xAI and DeepSeek. Microsoft said that the new AI medical tool correctly diagnosed 85.5 per cent of cases, which is way better compared to experienced human doctors, who were able to correctly diagnose only 20 per cent of the cases.
"This orchestration mechanism—multiple agents that work together in this chain-of-debate style—that's what's going to drive us closer to medical superintelligence,' Suleyman told The Wired.
Microsoft announced it is building a system designed to mimic the step-by-step approach of real-world clinicians—asking targeted questions, ordering diagnostic tests, and narrowing down possibilities to reach an accurate diagnosis. For example, a patient presenting with a cough and fever might be guided through blood tests and a chest X-ray before the system determines a diagnosis like pneumonia.
ALSO READ:
Sean Diddy Combs' secret plan against his ex Jennifer Lopez emerges amid sex-trafficking trial
Microsoft said its approach was able to wield a 'breadth and depth of expertise' that went beyond individual physicians because it could span multiple medical disciplines.
It added: 'Scaling this level of reasoning – and beyond – has the potential to reshape healthcare. AI could empower patients to self-manage routine aspects of care and equip clinicians with advanced decision support for complex cases.'
Microsoft acknowledged its work is not ready for clinical use. Further testing is needed on its 'orchestrator' to assess its performance on more common symptoms, for instance.
Economic Times WhatsApp channel
)
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Hindustan Times
27 minutes ago
- Hindustan Times
You won't believe how this teen became Microsoft's youngest cybersecurity star
What would you be doing right now if you were 13? Probably playing video games, scrolling through social media, or hanging out with friends. But a teenager was busy doing something quite extraordinary, finding security flaws in Microsoft's software and catching the company's attention at a very young age. Dylan, a school student, has become one of Microsoft's youngest and most celebrated security researchers, balancing studies and bug-hunting.(Microsoft MSRC) Microsoft recently featured Dylan in a company blog, calling him one of their youngest and most impressive security researchers. His journey with Microsoft started when he was just 13, collaborating with the Microsoft Security Response Center. Dylan's talent was so remarkable that Microsoft actually changed its Bug Bounty Program rules to allow 13-year-olds to participate, just so he could officially join their community. What makes Dylan's story even more remarkable is how early his passion for technology began. While most kids were still learning basic coding, Dylan was already exploring the source code of educational platforms in primary school, unlocking games on school computers and bypassing network restrictions during the lockdown. These early experiments might have landed him in trouble, but instead of misusing his skills, he chose a responsible path. He reported a major vulnerability he found in Microsoft Teams, which could let someone take over any group. That moment marked his entry into ethical hacking and responsible disclosure, a world where hackers help companies fix problems instead of exploiting them. Rising through the ranks Since then, Dylan's achievements have only grown. He has been named to Microsoft's Security Response Center Most Valuable Researcher list twice, in 2022 and again in 2024. Just last summer, he filed 20 vulnerability reports, more than triple his previous total. In April 2025, Dylan travelled all the way to Redmond, Washington, USA to compete in Microsoft's prestigious Zero Day Quest hacking event. Against some of the best minds in cybersecurity, he secured third place, proving that age is no barrier when it comes to talent and determination. Despite all these impressive milestones, Dylan still treats cybersecurity as a rewarding hobby. Like many school students, he balances his bug-hunting with studies, family time, and some sports. Microsoft's blog describes him as someone equally interested in science and civics, with plans to attend security conferences and learn from global experts when he's old enough. Dylan's journey is still unfolding, and it's hard not to wonder what he'll do next. For now, he's just a school student who happens to be on Microsoft's radar, juggling homework and hobbies while quietly making the digital world a little safer for everyone. Maybe next time you're sending a message or logging into a platform, you'll think of the teenager who helped keep things secure, all before finishing school.
Indian Express
32 minutes ago
- Indian Express
Uber, Ola can now charge twice the base fare: What surge, dynamic pricing means
The Ministry of Road Transport and Highways (MoRTH) announced on Tuesday (July 1) that cab aggregator companies such as Uber, Ola and Rapido may now charge customers up to twice the base fare as the maximum fare. The new Motor Vehicle Aggregator Guidelines, 2025, raised the previous limit of 1.5 times the base fare. 'The aggregator shall be permitted to charge a minimum of 50% lower than the base fare and a maximum dynamic pricing of two times the base fare…,' the guidelines said. Base fares are notified by the state governments for the corresponding category of motor vehicles. States have been advised to implement the revised guidelines within three months. What exactly is dynamic pricing, when do companies deploy the mechanism, and does it always prioritise drivers over customers? Is it the same as surge pricing? We explain. Companies resort to dynamic pricing models to benefit from the variations in the supply and demand of a product or service. This is, quite simply, market economics at work. Dynamic pricing follows the economic law of demand, which states that the price of a product will increase when its demand for an item or service increases in relation to its supply. This can happen due to a variety of factors. For example, in the case of cabs, rain can reduce the availability or the supply of cabs. The mechanism is widely used across several industries, including entertainment, utilities and hospitality. Amazon, for instance, has the dedicated 'Automate Pricing' tool, which allows sellers to set rules and parameters describing when and how prices may fluctuate for a range of products. However, the practice has also come under criticism. Last September, fans of the rock band Oasis held the ticketing platform Ticketmaster responsible for inflating concert ticket prices on checkout through dynamic pricing when the band announced its reunion tour in the UK. British Prime Minister Keir Starmer had termed the price hike 'depressing'. The Indian Railways first experimented with dynamic pricing in 2006. The then-Railways Minister Lalu Prasad proposed a dynamic pricing scheme for both freight and passenger trains, impacting peak and non-peak seasons, as well as premium and non-premium services. In 2016, it was formally adopted with the Railways adopting the 'flexi-fare' scheme. Under the scheme, the base fares of Rajdhani, Duronto and Shatabdi trains would increase by 10% with every 10% of berths sold, capping at 1.5 times the original fare. While the terms 'dynamic pricing' and 'surge pricing' are used interchangeably, surge pricing is simply one dynamic pricing strategy. It refers to the additional fee demanded from the consumer during periods of high demand, such as rush hours, when traffic congestion is at its peak. Such fees are also applied to utilities, like electricity, and are used to help manage the supply and demand, preventing potential blackouts. In the case of ride-sharing aggregators, the premise is simple: Rush hours correspond with periods of increased demand for rides from the likes of Uber, Ola and Rapido. With a larger proportion of users in an area demanding the cab service compared to the number of drivers available there, ride prices in the area surge. This presents the consumer with two choices — booking the ride at the higher price, or waiting for prices to reduce. The surge pricing also helps attract more drivers to an area to benefit from the increased prices, which ultimately helps reduce and normalise fares. According to the Uber website, its drivers are made aware of the surge price through a multiplier to the standard rate, an additional surge amount, or an upfront fare including the surge amount, which incentivises them to accept rides. The surge pricing can vary within the areas of a city. The Uber Driver app indicates the area-wise demand spikes through a colour-coded scheme. 'Light orange areas represent smaller earning opportunities from surge, while dark red areas indicate larger ones,' according to the website. In 2016, Wharton Business School researchers Gerard P Cachon, Kaitlin M Daniels and Ruben Lobel argued that the surge pricing model, while critiqued, did benefit Uber consumers by helping subsidise prices during off-peak times. Their research compared the surge pricing model to a benchmark fixed price and found that it also allowed Uber to serve markets that would otherwise remain underserved under a fixed pricing model, such as consumers hailing a cab on a rainy night. They also noted that Uber allows drivers to 'self-schedule', or decide their operating schedule, allowing them to gravitate towards areas facing a driver shortage when surge prices are effective. (Cachon, Daniels, Lobel, 'The Role of Surge Pricing on a Service Platform with Self-Scheduling Capacity', SSRN, 2016)
Time of India
42 minutes ago
- Time of India
OpenAI is Closing Down for one-full week
AI Image In what appears to be desperate bid to retain talent amid an aggressive poaching campaign by Meta, ChatGPT maker OpenAI has announced a mandatory week-long vacation for its employees. The move comes as the artificial intelligence (AI) race intensifies, with Meta luring at least eight OpenAI researchers to its newly formed "superintelligence" team over the past week, according to sources cited by Wired. This "recruiting coup," as described by the Wall Street Journal, has sparked a crisis at OpenAI, where a cloud of anxiety looms over the C-suite. The loss of talent is particularly stinging for OpenAI, a leader in AI research behind ChatGPT, as it faces mounting pressure to maintain its edge in the competitive AI landscape. Industry analysts note that this talent war reflects broader existential anxieties in the AI sector, where companies are racing to achieve artificial general intelligence (AGI)—AI capable of performing any intellectual task a human can. '80-hour work weeks' that make OpenAI leaders task more challenging Leaders are scrambling to keep staff loyal, a challenging task given that OpenAI employees often endure grueling 80-hour workweeks. The poaching has hit OpenAI hard, with insiders revealing a sense of betrayal among leadership. by Taboola by Taboola Sponsored Links Sponsored Links Promoted Links Promoted Links You May Like Kate Middleton Reportedly Taking Royal Split 'Badly' Crowdy Fan Undo What OpenAI Chief Research Officer Mark Chen told employees In a leaked internal memo posted to an OpenAI Slack channel, chief research officer Mark Chen expressed raw emotion, writing, "I feel a visceral feeling right now, as if someone has broken into our home and stolen something." Chen assured employees that he and CEO Sam Altman are working "around the clock" to counter Meta's aggressive recruitment tactics, which reportedly include staggering $100 million signing bonuses and first-year compensation packages. While OpenAI insiders confirmed these figures, some Meta sources disputed the numbers, according to Wired. One leader wrote, "If they pressure you, or make ridiculous exploding offers, just tell them to back off. It's not nice to pressure people in potentially the most important decision." The message, viewed by Wired, underscores the high stakes of the situation, with OpenAI framing Meta's tactics as aggressive and unethical. Meta's poaching spree is part of its broader pivot toward building human-level AI, an effort to recover from founder Mark Zuckerberg 's shifting priorities, which have included significant investments in the metaverse. One-week vacation and more: What OpenAI is doing to stop Meta from poaching its talent Chen's memo outlined proactive measures to retain talent, including recalibrating compensation and exploring "creative ways to recognize and reward top talent." Beyond the mandatory vacation, however, specifics remain vague. The week-long break, intended to help employees recharge, has raised concerns among OpenAI's leadership that Meta may exploit the downtime to intensify its recruitment efforts. One research lead warned in an internal message, "Meta knows we're taking this week to recharge and will take advantage of it to try and pressure you to make decisions fast and in isolation." The talent war has exposed deeper tensions. Seven other OpenAI research heads have also reached out to employees, urging them to resist Meta's overtures. AI Masterclass for Students. Upskill Young Ones Today!– Join Now
