A contemporary study conducted by researchers at Harvard University has revealed that advanced artificial intelligence systems are now capable of exceeding human doctors in both diagnosing medical conditions and determining treatment strategies, including in fast-paced and high-stakes emergency room environments. The research specifically accentuates the potential capabilities of modern AI systems in handling complex clinical reasoning tasks that were traditionally considered exclusive to trained physicians.
The findings, published in the peer-reviewed journal Science, are based on a controlled comparison between OpenAI o1 and experienced attending physicians. To ensure realistic testing conditions, the study used 76 actual emergency department cases sourced from Beth Israel Deaconess Medical Center. These cases were evaluated across multiple stages of the diagnostic process, allowing researchers to assess performance under varying levels of available patient information.
At the earliest stage of patient assessment, commonly referred to as initial triage, where clinicians typically have only limited details about a patient’s condition, the AI model demonstrated a notable advantage. It was able to correctly identify either the exact diagnosis or a closely related condition in 67.1 percent of the cases. In comparison, the two physicians involved in the study achieved accuracy rates of 55.3 percent and 50 percent respectively. This suggests that even with minimal data, the AI system was more effective at narrowing down potential diagnoses.
As the diagnostic process progressed and additional clinical information became available during the emergency room evaluation phase, the model’s performance improved further. Its diagnostic accuracy increased to 72.4 percent, reflecting its ability to refine its conclusions with more context. The physicians also showed improvement at this stage, but their accuracy remained lower, at 61.8 percent and 52.6 percent. This stage is particularly important as it mirrors real-world conditions where doctors continuously update their assessments based on new findings.
In the final phase of care, when patients were admitted either to general hospital wards or intensive care units, the AI model continued to outperform its human counterparts. It achieved an accuracy rate of 81.6 percent, compared to 78.9 percent and 69.7 percent for the physicians. Although the performance gap narrowed slightly at this stage, the AI still maintained a measurable edge, indicating consistency across the full diagnostic timeline.
Beyond identifying illnesses, the study also evaluated how effectively the AI system could design clinical management plans. This included decisions such as selecting appropriate medications, including antibiotics, as well as handling complex and sensitive scenarios like end-of-life care planning. Across five evaluated case studies, the AI achieved a median performance score of 89 percent. In contrast, physicians scored significantly lower, averaging 34 percent when relying on traditional clinical resources and 41 percent when supported by GPT-4. This underlines a substantial gap in structured decision-making support.
The researchers acknowledged that while integrating AI into clinical workflows is often viewed as a high-risk approach due to patient safety concerns, its potential benefits are significant. They noted that wider adoption of such systems could help reduce diagnostic errors, minimize treatment delays, and address disparities in access to healthcare services. These factors collectively contribute to both improved patient outcomes and reduced financial strain on healthcare systems.
At the same time, the study emphasizes that current AI systems are not without limitations. Clinical medicine involves more than text-based data. Doctors routinely rely on non-verbal and non-textual cues, such as observing a patient’s physical discomfort, interpreting imaging results, and making judgment calls based on experience. These aspects are not fully captured by existing AI models, which means human expertise remains essential.
The authors further concluded that large language models have now surpassed many traditional benchmarks used to measure clinical reasoning abilities. However, they stress the urgent need for more detailed research, including real-world clinical trials and studies focused on human-AI collaboration, to determine how these systems can be safely and effectively integrated into healthcare settings.
In comments shared with The Guardian, lead researcher Arjun Manrai clarified that the findings should not be interpreted as suggesting that AI will replace doctors. Instead, he described the results as evidence of a major technological shift that is likely to transform the medical field in the coming years.
From a macro industry perspective, this study reflects a developing trend in which AI is increasingly being used to augment clinical decision-making. However, experts continue to caution that challenges such as data bias, accountability, regulatory oversight, and patient trust must be addressed before such systems can be widely deployed. The future of healthcare, therefore, is likely to involve a collaborative model where AI amplifies efficiency and accuracy, while human doctors provide critical judgment, ethical oversight, and patient-centered care.
The RuÄ‘er BoÅ¡ković Institute (RBI) in Zagreb — Croatia’s biggest science and technology research center has confirmed it was one of thousands of organizations worldwide targeted in a massive cyberattack exploiting Microsoft SharePoint’s “ToolShell” security flaws.
The incident occurred on Thursday, July 31, 2025, and resulted in ransomware being installed on parts of the Institute’s internal network. According to RBI’s statement, the affected systems were linked to its administrative and support operations, with attackers encrypting documents and databases to block access.
Refusing to Pay the Hackers
Unlike some victims, RBI has stated it will not pay the ransom. Instead, the Institute plans to follow strict security protocols, restore affected systems from backups, and upgrade its infrastructure to meet modern cybersecurity standards.
Past reports indicate that ToolShell vulnerabilities have been used to spread two strains of ransomware — Warlock and 4L4MD4R but RBI has not yet confirmed which variant hit its systems.
Restoration Underway
Recovery work is ongoing, with some systems already back online. Email services were restored the Friday after the attack, and the Institute is slowly bringing other parts of its network back into operation. A completely new IT system is also being built to improve defenses and reduce future risks.
The response involves not just RBI’s internal team but also the Ministry of the Interior, Croatia’s national CERT, and other cybersecurity agencies. A detailed forensic investigation is still in progress.
Possible Data Exposure
It’s still unclear whether the attackers accessed personal information. Croatia’s Personal Data Protection Agency has been notified, and the Institute has pledged to act in line with GDPR rules if any breach of personal data is confirmed.
As a precaution, RBI’s data protection officer has already warned staff that some sensitive information, such as personal ID numbers, addresses, financial reimbursements, and other records may have been stolen. Employees were advised to stay alert for phishing emails pretending to be from the Institute or official authorities.
Part of a Global Problem
RBI is one of at least 9,000 institutions worldwide affected by attacks using the same ToolShell vulnerabilities. These flaws in Microsoft SharePoint have become a major cybercrime tool, enabling hackers to infiltrate networks, steal or lock data, and demand large ransom payments.
While the Institute continues its recovery, the attack is a reminder that even highly respected research organizations can be vulnerable, and that refusing to pay ransom demands can be both a security stance and a financial gamble.
Scientists at Sandia National Laboratories have achieved a significant milestone by developing ultra-compact optical chips that power quantum navigation sensors. These sensors utilize atom interferometers, a sophisticated technology that measures the interference patterns of atoms to track position and motion with unparalleled accuracy. Unlike traditional GPS, which relies on satellite signals, quantum navigation sensors operate independently, immune to external disruptions.
At the heart of this innovation lies the principle of quantum mechanics. Atom interferometers work by cooling atoms to near absolute zero temperatures, creating a state where they exhibit both particle and wave-like properties. When these atoms are subjected to laser pulses, they form interference patterns that can be precisely measured. By analyzing these patterns, the sensors can precisely determine changes in position and velocity.
The optical chips developed by Sandia National Laboratories are designed to be ultra-compact, making them suitable for integration into various devices and systems. These chips are capable of maintaining the delicate quantum states of atoms, ensuring accurate measurements even in challenging environments.
The potential applications of quantum navigation are vast and transformative. One of the most significant advantages is its ability to function in GPS-denied areas. This is particularly crucial for military operations, where GPS signals can be jammed or spoofed by adversaries. Quantum navigation ensures that military personnel and autonomous vehicles can navigate accurately without relying on external signals.
In addition to military applications, quantum navigation holds promise for the commercial sector. Autonomous vehicles, such as drones and self-driving cars, can benefit from this technology by achieving precise navigation in urban environments where GPS signals are often weak or obstructed. Furthermore, quantum navigation can enhance the accuracy of scientific research, particularly in fields like geology and archaeology, where precise location data is essential.
While the potential of quantum navigation is immense, there are challenges to overcome before it becomes mainstream. One of the primary challenges is the complexity of maintaining quantum states in real-world conditions. The ultra-cold temperatures required for atom interferometers are difficult to achieve and maintain outside of laboratory settings. However, the development of ultra-compact optical chips is a significant step towards addressing this challenge.
Another challenge is the integration of quantum navigation sensors into existing systems. This requires advancements in both hardware and software to ensure seamless compatibility. Researchers are actively developing robust algorithms and interfaces to facilitate the integration process.