Artificial intelligence is not new. In fact, its origins go back to ideas that emerged even before the first modern computers. But what we have today seems nothing short of magical.

AI chatbots can answer questions, write stories, help doctors detect diseases, perform a wide range of tasks, and even suggest what you should have for dinner.

But do they actually understand what they're saying? Or are they simply recognising patterns and generating responses that sound convincingly human?

In this episode of Talking of Science, Helen and Dr C are joined by machine learning engineer Kübra Altinel to unpack one of the biggest technological questions of our time: How can intelligence be artificial?

Together, they explore what AI actually is, how chatbots like ChatGPT generate answers, why they sometimes "hallucinate" information, and what makes them seem so convincingly human. Along the way, Kübra explains why AI is less like a person who understands the world and more like someone who has read millions of books but never stepped out of their home.

The conversation also explores how AI is already helping scientists and doctors, why probability lies at the heart of every chatbot response, and why healthy scepticism is one of the most important skills we can develop as AI becomes part of everyday life.

As always, Dr C's curiosity steers the discussion into unexpected places, proving that any complex topic can be understood through simple examples and analogies involving cakes, cookies, phone books, and grandmas.

To get in contact with us at Talking of Science, email hello@talkingofscience.dk.

To connect with Kübra Altinel, find her on LinkedIn.