Machine Learning Street Talk (MLST) is run by Tim Scarfe, Ph.D., and features regular appearances from Keith Duggar. It publishes weekly (15 episodes in the last 90 days, as of October 2026), and episodes usually run about 79 minutes.
The feed describes discussions with pre-eminent figures in the AI field, covering current affairs in AI, cognitive science, neuroscience and philosophy of mind, with the hype removed. It also states a belief in intellectual diversity in AI.
Recent guests and topics include Akarsh Kumar on evolution, Shawn Wen of PolyAI on voice agents, Leo de Moura on proofs, Zhengyao Jiang of Weco on AI research and Frank Hutter on tabular foundation models. The YouTube channel's most-viewed videos are very long conversations, several over two hours.
Our take: MLST suits listeners with a technical or academic interest in AI who want long, detailed conversations rather than news. Plan for roughly an hour and a half per episode. See our best machine learning podcasts list.
Ranked by YouTube views among the channel's latest 268 uploads, as of October 2026.
Built This Week is our show, with Jordan Metzner and Sam Nadler: the hosts build something with AI on air every week, test new AI tools, interview a founder or operator and cover AI news. MLST is a long-form research show; we are shorter and practical. For more picks, see best machine learning podcasts and best AI podcasts or the full list of best podcasts.
Show details on this page come from the public listings and feeds below, checked on October 24, 2026.
MLST is run by Tim Scarfe, Ph.D., with regular appearances from Keith Duggar.
Weekly. The feed shows 15 episodes in the last 90 days, as of October 2026.
Episodes usually run about 79 minutes.
By YouTube views, the top video is "The Ghost in the Machine — Noam Chomsky" (952K views as of October 2026), followed by "This is why Deep Learning is really weird."
The TWIML AI Podcast, Dwarkesh Podcast, Lex Fridman Podcast and Data Skeptic also feature in-depth AI and ML conversations.
