The Philosophical Snake: AI, Robotics, and a Fossilized Surprise

Episode Notes

The episode opens with the news that philosopher Daniel Dennett has died, and the hosts reflect on how influential his books, especially Darwin's Dangerous Idea and Consciousness Explained, were on Andrew's thinking about arguments, thought experiments, consciousness, and where the boundaries of sentience may lie. Brian adds his own examples from dogs and consciousness, reinforcing the sense that Dennett was especially good at exploring philosophical borderlands without always forcing tidy conclusions.

The middle of the episode is a long riff that starts with dog-powered historical machines and novelty inventions, then moves into Meta's Llama 3 release and a humorous discussion of Mark Zuckerberg's new public image. The conversation then turns to Boston Dynamics' new Atlas robot, broader robotics manufacturing and patent issues, cheaper robot dogs and quadrupeds from Unitree, and finally a fossil snake discovery from India that may have been about 50 feet long. The episode closes with pick discussions, including Andrew recommending Consciousness Explained and Justin recommending Shogun, while the later discussion also covers Never Split the Difference as a book they found interesting.

Key topics

• Daniel Dennett and consciousness: The hosts discuss Dennett's death, Andrew's admiration for his writing, and his use of examples involving the soul, sentience, severed limbs, and gradual brain replacement to explore consciousness and unclear boundaries.
• Dogs as workers and dog-powered machinery: Brian recounts a trainer's view that dogs want jobs, and Andrew uses that idea as a springboard into historical dog-powered machines, including treadmills, sewing machines, and other novelty contraptions.
• Old novelty catalogs and absurd inventions: The hosts joke about catalog-era inventions such as a 40-foot UFO and a dog-powered midget auto, reading a mock vintage description and comparing it to modern flim-flam and marketing nonsense.
• Meta Llama 3 and AI model evaluation: Andrew describes Llama 3 as a high-quality open-source model that appears competitive with GPT-4 on some evals, while noting that model comparisons are increasingly subjective because different models do different things well.
• Fast AI deployment and Meta's strategy: The hosts discuss how quickly AI features and models are being deployed on Meta's tools and platforms like Replicate, and Andrew frames this as evidence of Meta's serious but strategic open-source AI push.
• Boston Dynamics Atlas and robot manufacturability: The new Atlas robot is discussed as an impressive prototype, but Andrew emphasizes that the hard part is mass production at a price customers can accept, contrasting Boston Dynamics with cheaper competitors and referencing Rivian and Apple as supply-chain examples.
• Robot balance, servo control, and signal processing: Andrew explains that upright robot motion depends on computation, servo positioning, sensor feedback, and fast signal processing, connecting robot balance to the same sort of advances that made drones possible.
• Robot patents and defensive patent portfolios: The hosts talk about Figure AI, a gimbal-like joint design, and the use of patents as a defensive tool in crowded robotics markets rather than as proof of a finished product.
• Unitree robot pricing and consumer robotics: The discussion compares the expensive Unitree H1 and B2 robots with the much cheaper Go2 robot dog, using those prices to argue that robotics is moving toward broader accessibility and mass-market use.
• Fossil snake discovery from India: The episode ends with a fossil snake from India, reportedly from about 47 million years ago and possibly around 50 feet long, which the hosts describe as both cool and terrifying.
• Negotiation tactics in Never Split the Difference: The pick discussion centers on the hostage-negotiation book Never Split the Difference, especially open-ended questions like 'How am I supposed to do that?' and the distinction between real yeses and counterfeit yeses.

Picks

• Andrew Mayne: Consciousness Explained — Andrew explicitly recommends it as a good, accessible introduction to Dennett.
• Andrew Mayne: River Out of Eden — Andrew names it as one of his favorite books and describes it as a concise distillation of Dawkins's ideas.
• Justin Robert Young: Shogun — Justin clearly makes this his pick again and says he is still enjoying it and looking forward to the finale.