AI chatbots can now pass medical exams with ease and millions of people are already turning to them for health advice. But a new study from GP and clinical lecturer Rebecca Payne suggests that when it comes to real-world decisions, they may not be helping at all. She and colleagues found that people using leading AI tools were no better at identifying medical conditions or deciding where to seek care than those going it alone. So, if these systems know so much, why aren’t they better at helping? And what role should they really play in healthcare?

As countries scramble to secure reliable energy supplies, a lesser-known renewable is gaining attention beneath the waves. Unlike wind and solar, tidal power is predictable years in advance. Researchers have mapped more than 400 potential sites worldwide, revealing a surprisingly concentrated and powerful resource.

And debates about online safety often focus on teenagers, but what if older adults are just as vulnerable to misinformation, or perhaps even more so? Research suggests over-65s are far more likely to share fake news but very few receive any training to spot it. New evidence explores what training actually works but also reveals an unexpected risk - some approaches may make people more sceptical of everything, including the truth itself.