I am using AI to drop hats outside my window onto New Yorkers

Original Article ↗ Hacker News Discussion ↗

Overall Reaction

  • Many commenters find the project delightfully absurd and “atypically dumb in a great way,” praising the creativity, humor, and detailed writeup.
  • Others see it as overhyped: technically simple, framed with a clickbaity title that implies more autonomy and precision than exists.

How the System Actually Works

  • The setup uses a Raspberry Pi, stepper motor, yarn, and a computer-vision model (via Roboflow) to detect a person standing in a marked spot and trigger a drop.
  • It does not reliably place the hat on a person’s head; in the video the hat lands nearby on the sidewalk.
  • The service is opt-in: people book a time slot, pay, stand in a specific location, then receive a dropped hat.

AI vs. “AI” Debate

  • Several comments question whether this is really “AI” versus conventional computer vision (e.g., OpenCV).
  • Others argue image recognition and object detection are legitimately AI, and note the historical “AI effect” where yesterday’s AI becomes “just algorithms.”
  • Technical subthread discusses Roboflow, on-device vs hosted inference, Pi performance limits, and alternatives like Frigate/DOODS.

Safety, Legality, and Misuse

  • Concerns raised: potential injury from falling objects, distraction to drivers, risk to infants or vulnerable pedestrians, and general liability in a dense city.
  • Some argue this is akin to existing risks (people already can drop or throw things) and that intent and negligence matter more than the tech.
  • A darker line of discussion extrapolates to weaponization (grenades, bombs, drones), while others dismiss this as exaggerated.

Practicality and Business Viability

  • Skeptics doubt scalability or income potential; reloading and narrow location constraints limit throughput.
  • Others note the project functions more as art/marketing/whimsy than a serious delivery platform.

Related Ideas and Extensions

  • Many propose variants: balcony bead-throwers, lunch or gum drop services, pet feeders, sports/player tracking, face-tracking fans, vending-machine analogies.
  • Some compare it to earlier parachute/“jafflechute” drop concepts.

HN Meta-Discussion

  • Thread devolves at times into debates over puns, “fun vs. usefulness,” risk tolerance, and shifts in HN culture toward or away from playful hacker projects.