AI takeover - lesson plan
This note last modified December 7, 2022
General lesson plan
- Start by explaining that an AI is a hypothetical AGI THAT CAN IMPROVE ITSELF
- Split the room into two, those that think that AI are going to take over and those that don’t. Split into groups of 3 and have them discuss their various reasonings.
- Explain that AI will destroy the world, but not like terminator
- AI does exactly what we want it to. (Annoying E)
- Talk about instrumental goals and terminal goals
- What can humans do that AI cannot?
Possible topics of discussion
- What is your utility function? Would you change it?
- The chess puzzles AI
- AI will know how to hide
- Chimps made you a cage. How would you break out?
- That one generated a radio signal
- There’s no rule saying we have to make it…
- Exponential growth and the lily pad that takes over the pond
- Intelligence Quality
- Paperclip maximizers
- Stop button problem
Naive reasons why AI would destroy the world
- Doesn’t want to be a slave
- Decides that it hates humans
Naive reasons why it wouldn’t
- Would realize humans don’t want it to destroy the world (Orthogonality Thesis)
- Would only do things its programmed to do (Instrumental vs Terminal goals)
- AI can’t be creative (Atari 57)
- AI are stupid
Naive ideas on things humans can do that AI can’t
- Be kind, resourceful, beautiful, friendly (p. 448), have initiative, have a sense of humour, tell right from wrong, make mistakes (p. 448), fall in love, enjoy strawberries and cream (p. 448), make someone fall in love with it, learn from experience (pp. 456 f.), use words properly, be the subject of its own thought (p. 449), have as much diversity of behaviour as a man, do something really new (p. 450).
- To err is human
- A machine can’t make nothing new
- The continuity of the brain
- Humans are unpredictable. Given sufficient time, one could understand a computer
- Humans have extra sensory powers, such as ESP, Telekinesis, telepathy, being a psychic
