On this episode of Virtual Sentiments, Kristen Collins talks with Katharine Jarmul about how transparency, open-source AI, and privacy intersect in today's debates over artificial intelligence. Together they unpack the differences between proprietary, open-weight, and truly open-source AI models. They also explore how AI companies shape user experiences through design choices that often remain invisible, why smaller task-specific models may offer important advantages over increasingly centralized frontier models, and what concepts like data sovereignty and individual autonomy mean in practice. Rather than framing AI as an inevitable force beyond public influence, Collins and Jarmul make the case that curiosity, transparency, and community participation can help people reclaim agency over the technologies shaping their lives.

Katharine Jarmul is a data scientist who focuses her work and research on privacy and security in data science, deep learning and AI. She is the author of the well-received O'Reilly book, Practical Data Privacy, and has more than 10 years’ experience in machine learning and AI, where she has helped build large-scale AI systems with privacy and security built in.

**This episode was recorded on June 24, 2026**

Show Notes:

If you like the show, please subscribe, leave a 5-star review, and tell others about the show! We're available on Apple Podcasts, Spotify, Amazon Music, and wherever you get your podcasts.

Follow the Hayek Program on Twitter: @HayekProgram

Learn more about Academic & Student Programs

Follow the Mercatus Center on Twitter: @mercatus

Rss
Podden och tillhörande omslagsbild på den här sidan tillhör Mercatus Center at George Mason University. Innehållet i podden är skapat av Mercatus Center at George Mason University och inte av, eller tillsammans med, Poddtoppen.