In episode 26, the hosts continue last week's experiment of testing Claude AI's sense of humor, this time picking up a "Jean-Claude" naming joke with the newer Fable model and comparing how different Claude versions (Haiku, Sonnet, Opus, Fable) handle wit, speed, and depth. They dig into how Claude's model tiers are named and what that implies about capability versus response time. The conversation shifts to AI-generated content in the wild, from a viral "AI influencer" who reports on historical events as if she's really there, to deepfake-style clips reimagining world leaders in anime style. That leads into a discussion of "uncanny valley" and how hard it's become to tell AI-generated people, music, and art apart from the real thing, including a personal story about unknowingly enjoying AI-composed music on Bandcamp and YouTube. The hosts wrap up by putting Claude to the test with a dad joke request and a current-events question about the U.S.-Iran conflict, noting its surprisingly nuanced, source-backed response. They close by teeing up a question for next episode: just how powerful will AI get in the next two years?