Chinese researchers from Tsinghua University have developed “Agent Hospital,” an AI-powered virtual hospital town similar to Stanford’s AI town, which gained widespread attention last year. The innovative system features AI doctors and nurses autonomously treating virtual patients, with the goal of improving medical expertise through continuous evolution.

The AI agents, powered by large language models (LLMs), autonomously interact within the virtual hospital environment. According to the researchers, the AI doctors can treat up to 10,000 patients in a few days, a task that would take human doctors at least two years. These AI doctors have demonstrated impressive accuracy, achieving a 93.06% success rate on the MedQA dataset, which involves the full process of patient care, from diagnosis to follow-up.

At the National Business Aviation Association event in Orlando, Florida, Embraer introduced a revolutionary concept aircraft that offers a fully autonomous flying experience. This futuristic design showcases the potential of artificial intelligence (AI) in aviation, giving passengers a glimpse of what air travel could look like in the near future.

The aircraft features a cabin divided into three distinct zones, including a forward lounge for relaxation. One unique feature is the option for passengers to sit in the cockpit—if they wish—despite the plane being entirely operated by AI. Adding to its high-tech appeal, the windows are equipped with integrated touchscreens, enhancing the user-friendly interface.

Picking out individual ingredients in a dish can be a fun yet challenging part of enjoying a meal. While professional chefs and food scientists spend years honing their palates, a robot may soon be able to join in thanks to an AI-powered robotic taster developed by researchers at Penn State. Combining artificial intelligence with an electronic tongue, this advanced system can detect subtle differences in flavor and freshness that even human senses might miss.

In a recently published paper, the research team outlines how this AI “brain” works with an electronic tongue to analyze food at a molecular level. The robotic taster can distinguish minute variations in a cup of milk’s water content, identify the blend of beans in a cup of coffee, and even detect early signs of rot in fruit juice—well before human senses could.

For the first time, people can engage in a conversation with an older version of themselves—discussing life goals and aspirations—using an advanced AI-powered chatbot, complete with a photographic avatar of their future face. This innovative experience is part of the Future You project, created by researchers at MIT Media Lab in collaboration with international partners. The project aims to create a deeper connection between individuals and their future selves, using AI to simulate what users might look and think like when they’re 60 to 70 years old.

In a paper published on Oct. 1 to the preprint database arXiv, the researchers outlined the technology behind Future You. Through a text interface, users interact with an AI-powered version of their future selves on topics such as how to achieve life goals, considering their current circumstances and beliefs. The platform provides a way for individuals to feel more anchored to their future selves, as opposed to speaking with a generic AI chatbot.

In late September, Shield AI co-founder Brandon Tseng confidently asserted that fully autonomous weapons—where AI makes the final decision to kill—would never exist in the U.S. “Congress doesn’t want that. No one wants that,” Tseng told TechCrunch. However, his statement was quickly challenged. Just five days later, Anduril co-founder Palmer Luckey expressed a different perspective, signaling a more nuanced view on autonomous weapons.

Speaking at Pepperdine University, Luckey questioned the opposition to such systems, arguing that in some cases, autonomous technology may offer a moral advantage over current weapons. He raised the example of landmines, which can’t distinguish between a school bus and a military target, as a point of comparison. When asked for further clarification, an Anduril spokesperson noted that Luckey’s concern was more about the potential misuse of AI by bad actors, rather than advocating for fully autonomous lethal systems.

How do machine learning models work? And are they really capable of “thinking” or “reasoning” in the same way humans do? This philosophical and practical question has been the subject of ongoing debate. However, a recent paper titled “Understanding the Limitations of Mathematical Reasoning in Large Language Models,” authored by a team of AI research scientists at Apple, suggests a clear answer: not yet.

The core of the research centers around the difference between symbolic learning and pattern reproduction, and while these concepts are complex, the basic premise is straightforward. To illustrate, imagine a simple math problem: