AI Safety Issue important

ChatGPT Use cases and examples
1

Can someone please let me know where I can post to OpenAI my five years of research on critical AI safety issue? Thanks.

Thought Experiment: A Simple Game That Reveals Fundamental AI Blindness

A Thought Experiment Through a Philosophical Lens

Superposition Checkers (SC)

This simple game is standard checkers with one twist: after any capture, both players must take momentary control of an opponent’s piece and move it to any open square. Play continues normally until the next capture.

Consider two fun games children can pay:

Checkers

  1. *Mastered in 1995, Solved by AI in 2007.
  2. AI commands of a loyal army.
  3. At any moment, AI knows exactly where it stands.
  4. The solution is its Google Maps with GPS directions to guaranteed success and never get lost.
  5. No AI required.

Superposition Checkers

Please assume the complete theoretical solution is right in front of AI.

  1. AI commands a disloyal army with hidden traitors.
  2. At any moment, AI knows exactly where it stands, but has no memory of how it got there, and no ability to strategically see ahead.
  3. When an opponent captures a piece, AI is forced to make a quantum leap into an entirely new landscape. Once again, perfectly aware of its location — but total amnesia about the path behind, and no visibility of what lies ahead.
  4. AI isn’t navigating a map — it’s drifting in a tidal wave of murky waters.
  5. No amount of computation can map what is unmappable by design.

SC is not about checkers — it doesn’t show what AI can’t do, but what it doesn’t understand it’s doing.

Your thoughts?

  • Complete Years of AI safety research and Python code available on GitHub.com MIT Licensed
1 Like