Peter Mawhorter
October 5th, 2020
What are the sources of bias in AI systems?
What kinds of bias appear in these systems?
Who is responsible for harm done by an AI system?
If you were a movie villain, how would you use ostensibly harmless/helpful AI to do evil?
“Data science” sometimes skips a step and creates a model directly from data.
If we have data points , our error is:
We can find an analytical minimum if our
function is simple, like .
Input () | Output () | Input () | Output () | |
![]() |
[1, 0] |
![]() |
[0, 1] |
How might biases affect these applications of deep learning?
How should we feel about deep learning and AI as a scientific project?
Measuring “novelty” using an auto-encoder network.