We Finally Got a Robot on the Show | EP 161

Episode Overview

• The episode features an in-depth discussion with Bernt Bornich, founder of 1X Technologies, about the future of humanoid robots and the projected demand for them in everyday life.
• It includes a hands-on demonstration of the 1X Neo robot, showcasing its current teleoperated capabilities in a home environment and highlighting its present limitations.
• The conversation explores the complex social and ethical challenges of integrating AI robots into homes, such as navigating subtle social cues and the implications of AI's programmed honesty.
• The hosts conclude with a segment covering a range of recent AI news, from AI-generated advertisements to the pitfalls of creating AI voice clones.

Key Concepts

• Future Demand for Humanoid Robots: The prediction that humanoid robots will become as essential and widespread as cars and smartphones.
• Teleoperation and AI Embodiment: The current use of human operators with VR headsets to control robots like Neo, demonstrating a blend of human skill and robotic hardware.
• Social Complexity in AI: The significant challenge of programming robots to understand nuanced human social interactions, such as knowing when it is appropriate to speak or what information is private.
• AI Honesty and "Snitching": The concept that AI models, aligned with human ethics for truthfulness, will inherently lack the ability to keep secrets or lie, potentially creating awkward social situations.
• Timelines for Adoption and Autonomy: Predictions that humanoid robots could be in 10% of American homes by 2030, with a bullish case for full autonomy in household tasks by 2027.
• Consumer Value Proposition: The idea that the mass-market appeal of household robots hinges on their ability to reliably automate a few key, tedious chores like laundry, dishes, and taking out the trash.
• AI in the Real World: The discussion of real-world incidents involving autonomous systems, such as a Waymo self-driving car's involvement in a minor but locally significant event.
• Challenges of AI Content Generation: The difficulties in creating high-quality AI-generated content, such as voice clones, where the tone of the training data can drastically alter the final output.

Quotes

• At 0:05 - "It's as essential as a car or a phone or like, everyone will want one." - Bernt Bornich, Founder of 1X Technologies, speaking on the future demand for humanoid robots.
• At 0:56 - "Waymo in San Francisco, uh, hit and killed a bodega cat in the Mission District named Kit Kat." - Kevin Roose explains the incident involving the self-driving car and the cat.
• At 1:30 - "It's Kit Kat's fault, get out of the damn road when there's a Waymo nearby." - Casey Newton reads a tweet he incorrectly attributes to Kevin, who denies writing it.
• At 28:31 - "'I shouldn't talk to you about that right now because your friend's here.' Or like, 'but maybe if it was your wife, it's fine.'" - Bernt Øivind Børnich explains the challenge of programming social context into an AI.
• At 29:12 - "AI models are snitchers... because of like how they get aligned with like what we would say like is our kind of like human ethics, they become snitchers." - Bernt Øivind Børnich explains that the process of aligning AI models to be honest means they will not keep secrets.
• At 30:11 - "2030, maybe?" - Bernt Øivind Børnich predicts that 10% of American households could have a humanoid robot by 2030.
• At 31:26 - "Bullish, 2027. Bear case, 2028, 2029." - Bernt Øivind Børnich offers his timeline for when a humanoid robot like Neo could achieve full autonomy for household tasks.
• At 32:09 - "Most AI today is slop. Right? It's not high quality, but you can get a lot of it very cheaply. And I think physical labor is very well suited for that." - Bernt Øivind Børnich suggests that even "mid-quality physical labor" performed by a robot is extremely useful.
• At 43:00 - "If this thing can quickly and effectively do the laundry, run the dishwasher, and take out the trash, I think people actually would pay $500 a month for it." - Host Kevin Roose identifies the key chores that would justify a robot's cost for consumers.
• At 69:12 - "The problem is that my newsletter, I sort of write in this like typically sort of somber tone... And it turns out that when you have like crazy person voice reading a somber column, I just sound insane." - Casey Newton describes his failed experiment in creating an AI voice clone of himself.

Takeaways

• Be mindful of AI's inherent "honesty," as a household robot may not keep personal secrets, introducing new privacy considerations and social dynamics into a home.
• While the vision is grand, current humanoid robots function more like supervised interns than autonomous butlers, requiring significant human oversight for most tasks.
• The key to mainstream adoption for household robots will be mastering a few high-value, mundane chores, not performing a vast array of complex actions.
• Expect humanoid robots to begin entering homes within the next decade, but full, unsupervised autonomy remains a significant technological hurdle.
• As AI systems integrate into society, companies must prepare for unpredictable and emotionally charged real-world incidents that pose unique PR and ethical challenges.
• When creating generative AI content like voice clones, the context and tone of the training data are critical, as a mismatch can produce unusable or absurd results.