The theme of this year's Webby Award-winning IRL podcast is how to put people over profit in artificial intelligence. Episodes spotlight people putting ideas on how to build more trustworthy AI into practice: by examining datasets, recognizing humans behind data work and content moderation, scrutinizing AI-generated art and music, and more.
(THE INTERNET | OCTOBER 10, 2023) – Mozilla’s newest season of the podcast IRL: Online Life is Real Life debuting today showcases people and projects around the world building responsible AI alternatives.
Through a five-episode arc, the podcast interviews technologists who are confronting the societal risks of AI with practices, products, and principles in the public interest. In each episode, listeners will hear from a variety of voices, ranging from renowned AI researchers and policy experts, to the unsung and often invisible tech workers who help train AI systems.
AI is booming and the podcast transports listeners to the heart of important discussions shaping its development and its regulation. In the first episode, the podcast jumps right into the question of whether it's responsible to openly release large language models to the world.
Dr. Abeba Birhane, one of TIME Magazine’s 100 most influential people in AI and a cognitive scientist, speaks to IRL about her work auditing open datasets on internet content used to train AI. “The internet can be a really toxic place. It holds, you know, everything from the world's beauty to its ugliness and all, everything in between,” she says, explaining how reversing bias and hate speech in the outputs of large language models depends on openness.
Listeners are also transported to Nairobi, Kenya where over 150 content moderators of some of the most powerful tech companies in the world have gathered to form a union. In another episode, they hear from tech builders from Indigenous Māori communities in New Zealand reclaiming voice data and speech recognition for language revitalization.
Says J.Bob Alotta, Senior Vice President of Global Programs and Executive Producer of the IRL podcast, “What makes this season unique is how people across the globe are challenging the norm of profit first. We are using this formart to confront the crucial issues in building ethical AI, with solutions and optimism that AI can be trustworthy ”
Says Solana Larsen, editor of Mozilla’s IRL podcast: “Too often, it feels like we have lost control of the internet to the interests of Big Tech, Big Data — and now Big AI. In this season, we speak to trailblazers who are reclaiming power over AI to put people first.”
The launch of Season 7 coincides with Mozilla’s 25th anniversary. In 2023, IRL won the Webby for Top Technology Podcast.
Here’s an overview of the five episodes that will be released bi-weekly from October 10 to December 5:
With AIs Wide Open: AI builders on the risks and opportunities of open source large language models (like those used by ChatGPT) and their training datasets.
The Humans in the Machine: Data workers who are challenging tech companies, and entrepreneurs with new proposals for compensating the human labor behind AI.
Lend Me Your Voice: Language communities building AI speech recognition tools with their own data on their own terms worldwide.
Crash Test Dummies: Regulators and researchers on how to put the brakes on mass experimentation with AI systems on consumers and citizens.
The Art of AI: Hollywood writers, musicians and artists on setting new boundaries of consent when new art and music is generated by AI based on the old.
Europe: Tracy Kariuki | [email protected]
North America: Kevin Zawacki | [email protected]