This lecture is from the course Generative AI For All in Plain English - Essentials.
In this lecture, I explain the importance and complexity of bias in AI systems. I argue that defining what it means to be unbiased is challenging, if not impossible, and that people often want their own biases to be the ones used when claiming to be unbiased. We then explain why it is difficult to build unbiased AI’s because they are trained on inherently selective and biased data, resulting in the AI having biases similar to humans.
As a potential solution, I propose transparency about the biases of the AI’s we use, allowing people to make informed choices, even though it doesn't create unbiased AIs. I conclude by stressing the growing importance of this issue as AI becomes more prevalent in our lives.
