Schedule

Day 0. (Mon. Oct. 4th / Tue. Oct. 5th) Initialization:

You’re not going to do what you do without artificial intelligence.

Artificial intelligence is very important

We can’t ignore it, so we should get to know it

I want to help you know it as a friend

You will get to know it from a designer’s perspective

• Remind Sat/Sun students that there will be no class Sun. Oct. 24.

TODO’s

• Email new students
• Get professional Zoom link
• Reserve robot simulator time
• Put the schedule in my calendar
• Ask Dr. Park to understand if I do not arrive on time

Teaching

• Teach with care
  • Know their name
  • Let them know how much you care
    • Respect their learning preference: kinesthetic, visual, literate, etc
  • Get to know students’ interests and cater to each individual front row student
  • Respond to emails with “I’m sorry. I can’t reply now but I will get back with you by (state time)”
  • “Students are going to ask you about the larger questions of life. There are only a few of them that will go on to grad school and most of them will only remember what you taught them about anxiety or pressure and so you might be surprised in how you serve them.” (gradschool.usu.edu/grts/teaching-undergrads)
  • Help students get to know AI’s
  • Time 30 seconds for students to think of questions (you can’t judge 30s on your own)
• Get the students involved
  • Quick kahoots every so often (multiple times in a class)
  • Have them explain why they answered one way to their neighbor in 30s intervals
  • Warm them up for pair-explanation by group/table-explanation
  • Assure everyone “we want to hear your thoughts, questions, and comments”
  • Ask class for undivided attention for brief periods (even hold breath sometimes)
  • Ask questions to the degree they are willing to answer
  • Make sure the majority understand a topic before moving on
  • Keep it real: don’t spend too much time on abstract discussion
  • Remember that you’re just not going to get feedback from some students and forcing it will only make it worse
  • Find if some students would like to give an introduction to a topic
• Have fun!
  • kahoot!
  • Think up a lot of games to play
    • Have students throw marbles on unlevel landscape and try to get the lowest resting place
    • Also ask everyone to predict where the marble will roll (simulated because then you get the simulation bugs that don’t happen in the real world)
  • Give them real problems in lab to appreciate the need for more and more advanced architectures
  • Pass out worksheets: “what environment do I want to train my RL agent in” vs. “what skills do I expect my agent to develop”
  • Manually edit weights but you only get to see the loss derivative, activation value, and input value.
  • Students play GAN with their neighbors asking AI questions
  • Student vs AI.
  • Battle bots (family safe content; real and simulated)
  • Appreciate the need for MLops by asking teams to make high performance ANN's with no communication between the loss function (one student) and the ML engineering team (multiple students).
• Teach with enthusiasm
  • Use real examples to show how much you believe what you’re teaching
    • Robot will destroy something if the network doesn’t work correctly
  • end the lectures on a cliffhanger. get them wanting to come back
• Don’t teach yourself: only a handful of people are going to travels to the depths of AI that you have
  • If a student want to go in depth, save it for after class