What should drive the implementation of AI in legal services? I spoke on a panel about this question at the Recode AI & Law conference.
Here's my take:
➤ The implementation of AI in legal services should be driven by an acute awareness of the partiality of AI-based predictions.
➤ It is tempting to think of the predictions of AI systems as "objective" as they rely so heavily on mathematics, but they are not.
➤ While there is a growing awareness that bias in the data we feed to AI algorithms leads to biased predictions, this is only one of the reasons their predictions are socially and politically loaded.
➤ At the panel, I talked about how the use of accuracy for the evaluation of AI algorithms can lead to the reinforcement of social biases.
➤ No recording available, unfortunately