Introduction to AI

  • This lesson is specifically about machine learning.
  • Introduction:
    • Class discussion: Think-pair-share – what is artificial intelligence? Where have you seen it?
  • Inspiring with examples
    • Explain the story of the OpenAI Rubik’s cube solver, with the video playing in the background:
    • It’s not just amazing that it can solve Rubik’s cubes – computers could do that since just after the cube was released.
    • It’s not just amazing that it can solve it with a robotic hand – we’ve been using robotic machines since Unimate’s robotic arm in 1961!
    • What is amazing is that the computer could learn how to solve it in a simulated world, without ever seeing the real world. And when we put it in the real world, it could figure it out!
    • Hence, the incredible thing with machine learning is that it can learn to do things for itself, without being told how to do them.
    • Class discussion:
  • We can see the process of learning in Google’s “Agent57”, where it trains a computer program to outperform humans at 57 different Atari games
    • Again, it’s not impressive that computers can beat humans. Pong was made in 1972, and included an artificial intelligence enemy that could always beat the player.
    • What is impressive is that the algorithm can only “see” the individual pixels. It must learn what the up button does, what the down button does, how the ball moves – everything.
  • Although this is also very exciting, we can finish the lesson by a class discussion about ethics.
  • Case study: COMPAS (Correctional Offender Management Profiling for Alternative Sanctions )


+ There are no comments

Add yours