AI Ethics, Impossibility Theorems and Tradeoffs - Chris Stucchio | Crunch 2018

Channel:CraftHub Events · Length:43:23 · Published:2018-11-20 · Views:1605

Watch on YouTube

Summary

Chris Stucchio presents the ethical principles behind algorithmic decision making and shows with examples from lending, criminology and education that they cannot all be satisfied at once. The talk frames AI fairness as a set of quantitative tradeoffs.

Why HN recommends it

HN cites it in algorithmic-bias threads for the impossibility result that an algorithm cannot be procedurally fair, group fair and maximally useful at once.

it's impossible to simultaneously treat everyone the same way, to have a useful model, and to have uniformly distributed 'bias'-free outcomes. — HN, 2020

Machine learning & deep learning · AI safety, AGI & impact · Probability & game theory

Subtitles (excerpt)

[Applause] hey do we have slides cool so I'm the head of data science it's simple so as as mentioned we wrap a later platform which is kind of like a credit card and we're the largest one in India at the moment so since we're still a growing company I take on a lot of roles data science a lot of computer security and one of the other things I look at is ethical problems so that's what I'm gonna discuss here and in particular I'm gonna discuss about how ethical problems are difficult and there's no good way around them no matter depending on who you ask there's nothing you can do to avoid doing something that someone will consider unethical so they first introduce how the problem can arise so here's the simplest example of a machine learning algorithm that one can run and this is typically run by a store manager in a grocery store so you make a spreadsheet of items you sell and use and you calculate the shrinkage rate which is basically the rate at which people are stealing these products you sort this by how much money you're losing on each one which is the theft rate times the price and you put anti-theft devices on the ones that are high risk this should keep item so the anti-theft device is kind of an obnoxious plastic box that

On Hacker News

In many unbalanced situations you simply can't have an unbiased decision that's fair across all measures, no matter if the decision is done by a model, a human or a deity; you have to trade off between different types of unfairness. Equality of opportunity for individuals will result in unequal results for groups; equality of outcomes requires unequal opportunities if the historical circumstances… — PeterisP, 2021-02
I was interested in the slide deck and wanted to hear the author expand on some of his thoughts, found a video of his presentation if anyone else is interested. [VIDEO] — dvtrn, 2019-01
> I think it's also important to remember a lot of human models were abandoned because of the inherent bias The industry needs to stop misusing the term bias this way. Virtually every attempt to find this supposed human bias has failed. Latest public example was Amazon and hiring[1] Bias is the tendency to consistently mis-classify towards a certain class or tendency to consistently to over or un… — missosoup, 2020-01
This 'algorithm has bias' kind of headline needs to die. Every system has '''bias''' in the colloquial sense so long as the inputs to that system are not uniformly distributed. The only thing the designer of the system can do is make a conscious choice about which way to '''bias''' the system and be able to justify that choice. In practical terms it means that systems that deal with diverse popul… — missosoup, 2019-10
See also Stucchio's impossibility theorem: it is impossible for a given algorithm to be procedurally fair, representationally fair, and utilitarian at the same time: Video: [VIDEO] Slides: [LINK] — nostrademons, 2019-09

Related videos