Can artificial intelligence be a force for good?

Episode Summary

Executive Summary: This episode focuses on whether AI can be a force for good and what regulation is needed to make it safe. Shona McCallum interviews Gary Marcus, who argues for independent scientific oversight, pre-deployment testing, and stronger action against misinformation. The show also examines Wagner’s online influence in Africa, celebrates school AI innovators from the Isle of Man, and highlights efforts to bring more young women into engineering and electronics.

Main Topics: AI regulation and governance (Priority: 5/5): Gary Marcus argues that AI should be regulated through nimble national or global agencies with scientists and ethicists involved, rather than leaving rules to governments and companies alone. Misinformation and democratic risk (Priority: 5/5): The episode stresses the danger of AI-generated text, images, and voice clones making misinformation cheap, plausible, and scalable, with possible effects on elections and public trust. Global AI summit and international cooperation (Priority: 4/5): The upcoming UN-linked AI summit in Geneva is framed as a chance for countries, companies, scientists, and ethicists to agree on guardrails, including possible Chinese participation. Wagner and Russian influence in Africa (Priority: 4/5): BBC Monitoring analysis explores how Wagner-linked and pro-Russian messaging spread in African social media ecosystems, especially during the mutiny against Prigozhin, and how local influencers amplify it. Youth innovation in AI for road safety (Priority: 4/5): Team Apollo from the Isle of Man wins an international award for a low-cost AI heads-up display that warns drivers about speed, navigation, pedestrians, and vehicles. Women in engineering and semiconductors (Priority: 3/5): Report from Scotland and Africa highlights gender imbalance in engineering and semiconductor careers and describes outreach efforts to make STEM more accessible and practical for girls.

Key Arguments: AI should be regulated with independent scientific oversight before large-scale deployment, like pre-approval for medicine. A flexible AI agency is better than detailed legislation because the technology changes too quickly for static rules. Governments, companies, scientists, and ethicists should all have seats at AI governance tables to avoid regulatory capture. Countries have shared interests in stopping cyber scams, misinformation, and concentrated tech power, even amid geopolitical tensions. AI-generated misinformation is likely to heavily affect elections and could undermine democracy if trust in information collapses. Deepfakes and voice cloning expand scams, propaganda, and fraud beyond text alone. The greatest near-term AI danger is catastrophic misuse, not immediate extinction; risks include conflict, cyberattacks, and infrastructure disruption. Practical, hands-on education and better career visibility can help more girls enter engineering and semiconductor fields.

Data Points: Countries represented at UN AI summit: 183 - Expected representatives at the global AI summit in Geneva next month. Age of Team Apollo members: 17 - Danielle Wilkinson and Praveen Balakrishnan are both 17-year-old students. Project development period: 6 months - Team Apollo spread their AI car-safety project across about six months. Engineering workforce women share: 16.5% - Engineering UK figure cited for the proportion of engineers who are women. Semiconductor workforce women share: 20% to 25% - Global Semiconductor Alliance and Accenture estimate referenced in the report. AI age referenced by speaker: 75 years - Gary Marcus notes that AI has existed for 75 years, separate from deep learning. WhatsApp listener line: 0330-123-0320 - Contact number given for listener feedback. Email address: [email protected] - Contact email given for listener feedback.

Pivotal Quotes: "We need to require companies to do that, and it can't be just the governments who don't have the expertise, and it can't be just the companies that make those decisions." — Gary Marcus: On why independent scientists should help evaluate AI before wide deployment. "The cost of making misinformation really like fluent, plausible-sounding misinformation, has basically gone to zero." — Gary Marcus: On why AI-generated false content is now a major societal threat. "Driving safety should be for everyone and not just for those with certain cars." — Danielle Wilkinson: On Team Apollo’s motivation for building an affordable AI heads-up display.

Implications: The episode suggests AI needs urgent, science-led guardrails to limit fraud, deepfakes, and democratic harm while still enabling useful innovation. It also shows young people and underrepresented groups can shape safer tech if given access, support, and practical training.