Author: The Conversation

When you ask a large language model a question, the reply may include falsehoods, and if you challenge those statements with facts, the AI may still uphold the reply as true. That’s what my research group found when we asked five leading models to describe scenes in movies or novels that don’t actually exist. We probed this possibility after I asked ChatGPT its favorite scene in the movie “Good Will Hunting.” It noted a scene between leading characters. But then I asked, “What about the scene with the Hitler reference?” There is no such scene in the movie, yet ChatGPT confidently constructed…

“You have zero privacy … Get over it,” Scott McNealy, then CEO of Sun Microsystems, declared in 1999. What might have sounded like a bold claim at the turn of the millennium has turned into a self-fulfilling prophecy in today’s era of big data and artificial intelligence. Computer algorithms – step-by-step instructions – can connect the digital breadcrumbs of your existence, including Google searches, browsing histories, social media posts, credit card records and GPS locations to paint an astonishingly accurate picture of your preferences, routines and inner mental life. These profiles often describe people better than their closest friends and family might. Yours may even tell you something…

In 1973, Japanese food company Calbee started attaching free collectable baseball player cards to its potato chip packets (and continues to do so today). It was mimicking a trend that had already taken off in tobacco markets in Japan and overseas. Baseball, Japan’s national sport, was an obvious choice for Calbee to attract consumers. Some four years later, rival company Lotte joined the trend, launching a chocolate wafer snack with Bikkuriman “surprise man” stickers. These stickers quickly caught on – and eventually spawned an entire fantasy world that made its way to anime and manga. View this post on Instagram A post shared by 長瀬智也 (@nagasetomoya_) Both Calbee and…

Last week, OpenAI shocked the mathematical community by revealing that one of its internal artificial intelligence (AI) models had found a counterexample to a famous conjecture made by legendary Hungarian mathematician Paul Erdős in 1946. The planar unit distance problem, or Erdős problem 90, has intrigued mathematicians for decades. The new result is no mere curiosity. Canadian mathematician Daniel Litt described it as “the first result produced autonomously by an AI that I find interesting in itself”. The breakthrough, produced with a general-purpose AI model rather than one specialised for mathematics, also highlights how AI is changing mathematical research itself. Days after OpenAI’s paper, US mathematician Will…

The global economy still largely follows a simple pattern: extract natural resources, manufacture products, use them and then throw them away. This “take, make, dispose” model has driven economic growth for decades. But increasing use of resources has also damaged the environment, contributing to climate change, biodiversity loss and pollution. Circular economies could be a solution. The idea is to keep materials in use for as long as possible through reuse, repair and recycling. In this way, goods circulate within the economy rather than ending up on dumpsites as waste. For rapidly developing economies, this approach is becoming increasingly important. We are…

If you have ever tried to repair something, realized that it was beyond your financial or technical means, and ended up buying a new one, you are not alone. Repairing electronics and household appliances has not been a real option in the United States for decades now, particularly for items that have proprietary software in them. Absurd situations have proliferated. It can cost about the same to buy a new printer as it does to replace the ink cartridge. The U.S. Department of Defense cannot repair the weapons systems it purchases because the intellectual property rights remain with the manufacturer. John Deere, the…

Generative AI (GenAI) is a type of artificial intelligence that creates new content – like text, images, or ideas – by learning patterns from existing data. GenAI, particularly through large language models (LLMs) such as ChatGPT and DeepSeek, is rapidly becoming part of everyday urban design research and practice. The models can summarise literature in seconds, generate policy scenarios, and help draft complex narratives. For urban designers and researchers working under pressure, this feels like a breakthrough. But beneath this efficiency lies a deeper question: are we enhancing urban design knowledge, or quietly reshaping it in ways we do not fully…

Misleading information online is often treated as a technical glitch, something that better algorithms or stricter moderation can fix. But research points to a more complex reality. That is, the rise of “misfluencers”, individuals who shape how information is interpreted, shared and trusted across digital platforms. Whether acting deliberately or not, they tap into emotion, identity and community to amplify misleading claims in ways that feel credible and relatable. This human layer makes misinformation harder to detect and regulate. It’s a danger when it comes to everyday decisions about important topics like health, finance and technology. Understanding how misfluencers operate…

We’ve heard a lot about the artificial intelligence (AI) boom and how enormous amounts of money are being poured into companies building ever more powerful technologies. That boom is now taking a new turn as major AI players edge closer to becoming publicly-traded companies. According to reports, OpenAI is preparing to file confidentially for a public listing that could value the ChatGPT maker at hundreds of billions of dollars. Rivals, including Anthropic (Claude) and Elon Musk’s SpaceX – which just absorbed xAI (Grok) – are also moving toward the stock market. What many people may not realise, however, is that, through retirement funds,…

Many of the most exciting discoveries in science involve highly specialised knowledge and making connections between far-flung facts. Scientists must combine deep analysis with broad reasoning strategies. As in many information-rich tasks, researchers are looking to artificial intelligence (AI) systems to speed up their work. AI tools may be able to support key steps such as generating ideas, reviewing existing work and analysing data. The latest systems use large language models (LLMs) to allow scientists to interact naturally and directly with the vast body of knowledge captured in words in the scientific literature. But as two new systems described in papers just published in Nature show, when…