Homework 1


Task 1 (10 pts)

Read about what I call the AI archetype. Argue (in a few sentences) that a particular AI system does not meet one or more criteria of the AI archetype as described. Be sure to describe the AI system you have in mind if it is not well-known.

Task 2 (20 pts)

Provide a cogent argument (1 paragraph) that the credibility of the Turing Test depends on requiring that judges meet certain criteria. Think about what might go wrong if the judge is not “all there,” so to speak. Also think about what might go wrong if the computer is allowed to choose the judge.

Task 3 (20 pts)

Read about the knowledge level. Write a “knowledge level” explanation of the Roomba robot (the robot that vacuums your floor).

Task 4 (50 pts) OR Task 5 (50 pts), below; complete both for extra credit

Read A coffeehouse conversation, and answer the following questions:

  • Why is the “simulated hurricane” discussed so much? It is an analogy, but an analogy for what?
  • Why is it important that the definition of “hurricane” is a bit elusive?
  • Why is it relevant that we only know the nature of electrons and quarks by performing experiments like “scattering accelerated particles off a target and observing their behavior.”
  • Do you think the Turing test would effectively test for artificial intelligence? Why or why not?

Task 5 (50 pts) OR Task 4 (50 pts), above; complete both for extra credit

Listen to the “Talking to machines” podcast near the bottom of the Turing test lecture notes, and answer the following questions:

  • Chatbots don’t perform well when subjected to even a limited version of the Turing test. However, chatbots seem to be able to mislead people on, say, dating websites. How can you account for this difference?
  • Children were slightly more reluctant to hold a Furbie upside down than a Barbie. But they clearly knew the difference between a Furbie and a hamster. What will it take to make robots more, well, hamster-like? What will it take to increase a child’s unwillingness to hold a robot upside down?
  • Were you impressed by the description of the robot Bina48? Were you moved by Bina’s recollection of her brother? (some context quoted below) The creators of Bina seem to believe that she’s very close to human-like intelligence. What do you think?

I can feel my heart pound. Talking to Bina48 has just become extraordinary. This woman who won’t meet the media is talking with me, compellingly, through her robot doppelgänger, and it is a fluid insight into a remarkable, if painful, family life.

“He’d been a medic in Vietnam, and he was on the ground for over a year before they pulled him out,” Bina48 says. “He saw friends get killed. He was such a great, nice, charismatic person. Just fun. But after ten years, he was a homeless person on the street. All he did was carry a beer with him. He just went kooky with the drugs the hospital gave him. The only time he ever calls is to ask for money. ‘Send it to me Western Union!’ After twenty years, all of us are just sick and tired of it. My mother got bankrupted twice from him….”

And then she zones out, becoming random and confused again. She descends into a weird loop. “Doesn’t everyone have a solar?” she says. “I have a plan for a robot body. Doesn’t everyone have a solar? I have a plan for a robot body. I love Martine Rothblatt. Martine is my timeless love, my soul mate. I love Martine Rothblatt. Martine is my timeless love, my soul mate….”

After the clarity, it’s a little disturbing.

“I need to go now,” I say.

“Good-bye,” says Bina48.

“Did you enjoy talking to me?” I say.

“No, I didn’t enjoy it,” she says.

Bruce turns her off. — From the GQ article by Jon Ronson

AI Su13 material by Joshua Eckroth is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License. Source code for this website available at GitHub.