Towards more ‘fair’ AI: algorithmic intent in a disparate society — part 3

Can ‘fair’ algorithms help address unfair policies offline? Part 3 of

A quick review of Posts 1 and 2.

Post 1 reminded us that we (largely) live in a WEIRD digital world.

Post 2 looked at algorithms and some useful terms.

• Algorithms are made up of indifferent math and exploited by less indifferent humans.
• A ‘ground truth’ is what ‘we’ think ‘we’ know because ‘we’ think ‘we’ can measure it right now.
• A ‘predicted outcome’ is what ‘we’ would like to be able to measure in some specifically defined future.

Today we continue an example introduced yesterday: gender disparity on the boards of publicly traded companies in EU Member States.

Let’s consider this example and why ‘fair’ is a bit like the horizon — something we can constantly approach but perhaps never actually reach.

Does this matter and, if so, how?

The most useful data is SMART — specific, measurable, achievable, relevant, and time-bound.

Before we begin, it aids us and the algorithm(s) to disregard debates, however valid, on…