Brian and Andrew talk about… HOLY MOLY JUSTIN IS BACK!

Episode Notes

The episode opens with a discussion of Google's Gemini Ultra launch and the rebranding of Bard to Gemini. The hosts compare Gemini with ChatGPT on model quality, pricing, integrations, and usefulness for coding and summarization, while also talking about how AI tools are already woven into their workflows for transcripts, show notes, and quick summaries. Brian also describes using AI for a real-world well-and-solar calculation experiment, using photos and measurements to estimate energy use, water flow, and costs.

Later, the conversation turns to the joke AI model Goody2, which is framed as an exaggerated parody of overcautious safety behavior because it refuses nearly every prompt. The back half of the episode focuses on Sam Altman and the scale of future AI compute demand, including a reported $7 trillion chip effort, followed by a live discussion of Andrew's Apple Vision Pro demo, its comfort and display tradeoffs, security concerns about deepfakes, the promise of lighter AR glasses, and a round of media picks including The Greatest Night in Pop and In Deep Geek.

Key topics

• Gemini vs. ChatGPT feature comparison: The hosts compare Google's Gemini Ultra with ChatGPT in terms of model quality, coding utility, pricing, app ecosystem, and integrations like Gmail, Docs, Sheets, and search defaults.
• AI as a practical assistant for technical and household tasks: Brian's well/solar example shows AI being used for real-world estimation, electricity cost calculations, and determining how much a pond would take to fill.
• AI show-note and transcript workflows: Justin describes using Riverside, Whisper-style transcription, and summarization workflows to turn large recordings into transcripts and condensed show notes.
• Overly defensive AI safety behavior: Goody2 is presented as a deliberately ridiculous AI that treats every query as dangerous or offensive and therefore refuses to answer almost anything.
• AI infrastructure and semiconductor demand: The Sam Altman discussion centers on the need for massive amounts of compute and chips, with the hosts debating whether current and planned microprocessor supply can meet future AI demand.
• Electrification as an analogy for AI infrastructure growth: Andrew compares AI adoption to the historical rise of electrification, arguing that once AI is widely used for cognitive work, compute demand could rise even more steeply.
• Apple Vision Pro ergonomics and design tradeoffs: The hosts discuss the headset's weight, external battery, eye tracking, display quality, and the engineering compromises Apple made in order to keep it usable.
• Deepfake risk and secure identity verification: Andrew emphasizes the danger of voice and video cloning scams and notes that FaceTime and Apple’s iris scanning suggest a focus on trust and identity security.
• Everyday use of lightweight AR glasses: Andrew describes using Xreal-style glasses for video watching and phone mirroring, noting they are more normal-looking than Vision Pro but far less immersive.
• Apple Vision Pro as a premium viewing device: The hosts see the strongest current use of Vision Pro as media consumption, especially movies, panoramas, and 3D video, even if it is not yet an essential work device.
• Two-tier future for mixed reality and glasses hardware: Justin suggests the market may split between expensive serious headsets and lighter, cheaper devices for quick interaction, and Andrew agrees that lighter products will likely be more popular for simple tasks.
• Phone-dependent AI glasses and offloaded processing: The discussion covers inexpensive glasses that rely on a phone for power or processing and use AI APIs for things like translation and image generation.
• Skepticism about replacing phone apps with standalone AI assistants: Rabbit R1 and Humane are treated as examples of devices that promise to replace app-based smartphone workflows but may just create extra friction.

Picks

• Justin Robert Young: The Greatest Night in Pop — Strong recommendation; he says he really liked it, praises the footage and access, and calls it a great documentary about how We Are the World came together.
• Brian Brushwood: I Think You Should Leave on Tour — He explicitly calls it his pick and encourages listeners to watch related shows and consider seeing it live in Dallas or Houston.
• Andrew Mayne: In Deep Geek — He recommends the channel enthusiastically for deep Tolkien lore and fantasy deep-dives, calling the videos well presented and a delight.