Generative artificial intelligence on our “Babbage” podcast

GENERATIVE AI is taking the world by storm. Babbage has explored artificial intelligence through a variety of lenses.

Since the launch of OpenAI’s ChatGPT in November 2022, little else has occupied the minds of technologists. Generative artificial intelligence—capable of producing media like text, images and audio in response to prompts—seems to be improving every day, with many technology companies developing and releasing their own competing systems.

As the AI revolution accelerates, the technology is being used in ever more creative ways, companies are discovering its potential, causing unease among many content-creators and white-collar professionals, whose jobs seem to be at risk. The story of automation changing the world of work is not a new one. But the speed, the visibility and the hype surrounding generative AI can seem alarming. How worrying is it?

The Economist’s Abby Bertics and Arjun Ramani explain how large language models work, the risk posed by the technology—and what to do about it. Callum Williams, our senior economics writer, ponders the potential for economic disruption as generative AI enters the workplace. Plus, Tom Standage, The Economist’s deputy editor explores the question of regulating this emerging technology without hindering innovation. Kenneth Cukier hosts.

OpenAI’s ChatGPT, an advanced chatbot, amassed over 100 million monthly active users and exhibited unprecedented capabilities, from crafting essays and fiction to designing websites and writing code. You’d be forgiven for thinking there’s little it can’t do.

In March 2023, it had an upgrade. GPT-4 is even more impressive. It can take in photos as an input, and deliver smoother, more natural writing to the user. But it also hallucinates, throws up false answers, and remains unable to reference any world events that happened after September 2021.

Seeking to get into the nuts and bolts of the Large Language Model that operates GPT-4, The Economist’s science and technology editor, Alok Jha, speaks with Maria Laikata, a professor in Natural Language Processing at Queen Mary University of London. We also put the technology through its paces with The Economist’s tech-guru Ludwig Seigele.

How significant will GPT-4 and similar technologies be in the grand scheme of machine intelligence? Perhaps counter-intuitively, Gary Marcus, an expert in both AI and human intelligence, thinks their importance has been overstated—though they will impact all of our lives both in good and bad ways.

Engineers have spent decades trying to create functional humanoid robots, which look and act like humans. But these machines, which combine complex mechanics with generative artificial intelligence models, like ChatGPT, are finally coming-of-age. Are they good enough to sustain a human-like conversation, though?

Host Alok Jha travels to Cornwall to meet Ameca, a robot made by Engineered Arts. Will Jackson, the company’s boss, explains how the technology behind Ameca works and the advantages of having robots that look and behave like people. Plus, Paul Markillie, The Economist’s innovation editor, assesses the state of the field and how to prepare for the rollout of humanoid robots.

Digital fakery, from the latest generation of deepfakes to lower-tech trickery, threatens to erode trust in societies and can prevent justice from being served. But how can technology be used to both detect deepfakes and authenticate real images?

Patrick Traynor, a professor at the University of Florida, explains a novel method to expose audio generated by artificial intelligence. Ilke Demir of Intel Labs demonstrates how to spot visual fakery by analysing colour changes in the face. Plus, The Economist’s Benjamin Sutherland investigates the flipside of deepfakes: how to prove that footage is real. And Wendy Betts of eyeWitness to Atrocities explains how her technology is being used as evidence for war crimes. Alok Jha hosts.

ChatGPT is just one example of a new type of artificial intelligence, which could become the next major general-purpose technology. We explain why “foundational AI” promises to be so transformative.

The Economist’s Ludwig Siegele explains why ChatGPT has been taking the world by storm, and why foundation models, or generative AI, could end up having an economic impact similar to that of electricity. Jack Clark of Anthropic AI tells us about the new AI ecosystem that is emerging. Reeps One, a beatboxer, explains how he uses machine learning to compose music. And Kate Crawford, the author of “Atlas of AI”, considers why the technology is proving so controversial. Alok Jha hosts.

A debate has been raging in technology circles, after an engineer at Google claimed in June 2022 that the company’s chatbot was sentient. Host Kenneth Cukier explores how to define “sentience” and whether it could be attained by AI. If machines can exhibit consciousness, it presents myriad ethical and legal considerations. Is society equipped to deal with the implications of conscious AI?

Babbage from The Economist is released every Wednesday.

Listen on: Apple Podcasts | Spotify | Google | Stitcher | TuneIn

Find all of The Economist’s podcasts at economist.com/podcasts.