Stojnic et al (2022)

1. Describe the “violation-of-expectation” looking-time paradigm used in this study?

2. Infants performance on BIB
   1. Infants looked significantly longer in the unexpected condition in which of the six tasks (Figure 2, p. 7)?
      • Goal-directed
      • Multi-agent
      • Inaccessible-goal
      • Efficient-agent
      • Inefficient-agent
      • Instrumental action
   2. Of the tasks that were significant, explain what knowledge about agents they were meant to test?

3. Explain how the performance of the models differed from infants (Section: “Machine performance on BIB”).

4. What do you think about the approach of setting infant intelligence as a benchmark for AGI?

5. Do you think that AGI is a goal that researchers should be pursuing?

6. Do you have any questions that you want to discuss in class?