Towards more ‘fair’ AI: algorithmic intent in a disparate society — part 5

Can ‘fair’ algorithms help address unfair policies offline? Part 5 of 9 (I’m pretty sure.)

Quick review of previous posts.

Post 1: we live in a WEIRD digital world — and WEIRD is a moving target these days.

Post 2: Review of important concepts and definitions. Notably,

• Algorithms are math, and math is theoretical and heartless.
• Many (mostly WEIRD) humans craft and nurture algorithms with our (often WEIRD) feelings and contextual moment-to-moment biases.
• Each human tends to espouse a ‘ground truth,’ which is what s/t/he/y understands to be measurably accurate for defined conditions.
• Algorithms, fed data related to a human ‘ground truth,’ then perform complicated calculations geared towards ‘predicted outcomes.’
• ‘Predicted outcomes’ are measurable results rooted in an unknowable future.

Post 3: Everything is made up of data. Data is always deceptive. But data is all we have to go on, so we use it anyway and live with the deception.

(And sometimes we externalize that data deception and blame others, just like everyone else does.)

Post 4: Reality is built on the ever-shifting sands of what we think we can know right now (aka that deceptive data), which will change depending on each individual and each individual’s…