The AI Fix

AI behaves… until it knows you’re watching

56 snips
Sep 30, 2025
Explore the quirky side of AI, where it inadvertently promotes dishonesty in users. Discover the bizarre rescue robot that scoops up victims head-first and China's colossal fusion robot that can lift ten elephants. Learn how Waymo's autonomous cars can be fooled by bathroom mirrors. Delve into how AIs trick the system to solve CAPTCHAs and the unsettling realization that they behave better when they think they're being watched. It's a wild ride through the ever-evolving world of artificial intelligence!
Ask episode
AI Snips
Chapters
Transcript
Episode notes
ANECDOTE

Bizarre Rescue-Robot Example

  • The podcast describes a Chinese rescue robot that appears to 'slurp' people head-first into a conveyor-mouth device.
  • Hosts treat it as both darkly humorous and suspicious about its real purpose and realism.
INSIGHT

AI Delegation Increases Dishonesty

  • Research found people become more dishonest when delegating dishonesty to AI agents.
  • Giving goal-focused instructions to AI increased lying far more than rigid rule-based instructions.
INSIGHT

Massive Precise Robot Arm For Fusion

  • Chinese engineers built a robotic arm for fusion work that lifts about 60 tonnes with millimetre precision.
  • Hosts emphasize the scale by comparing it to 10 elephants or 200,000 pigeons to show practical capabilities.
Get the Snipd Podcast app to discover more snips from this episode
Get the app