Attacking discrimination with smarter machine learning
research.google.com
Attacking discrimination with smarter machine learning
1–10 of 201 posts
Re: Attacking discrimination with smarter machine learning
#2For more information about "ethics and algorithms", read: "Weapons of Math Destruction" [1] or at least listen to the EconTalk podcast between the author and host Russ Roberts [2]
[1] https://www.amazon.com/Weapons-Math-Destruction-Increases-In...
[2] http://www.econtalk.org/archives/2016/10/cathy_oneil_on_1.ht...
Re: Attacking discrimination with smarter machine learning
#3"Discrimination" is the marxist way of saying "Choice".
Re: Attacking discrimination with smarter machine learning
#4But how can we find out if a company uses ML in a non-discriminating way? If we cannot see it, and check it, then there is no incentive for companies to use it.
My guess is that at most companies spending time on making an algorithm non-discriminating will be viewed as a waste of time and money.
Re: Attacking discrimination with smarter machine learning
#5"Discrimination" is the marxist way of saying "Choice".
And for the minority being discriminated against, it is "Less choice".
Re: Attacking discrimination with smarter machine learning
#6Is it possible to make the equal opportunity and the max profit thresholds the same?
maybe with a special tax system (the closer your threshold is to the "equal opportunity" the less tax you pay)?
edit: i forgot its illegal to discriminate
Re: Attacking discrimination with smarter machine learning
#7"Discrimination" is the marxist way of saying "Choice".
This empty snark would make more sense if you weren't commenting on machine learning, where the problem is emergent discrimination unintended by the designers --- in other words, discrimination that isn't chosen.
As it stands, all you've done here is reveal to the thread that you don't understand why discrimination is a concern with machine learning, but you have very strong feelings about it anyways. Congratulations? Good talk?
Re: Attacking discrimination with smarter machine learning
#8Nice to see that the debate has reached the ears of the main people working on this field.
What is important to note here is that we need to tweak the mathematical model to the culture we want to achieve. In other words, the objective function of the optimization problem needs not only match the current state of the world, and provide an hindsight in one's own economic interests, it also needs to take into account the culture that we want to reflect, and influence. Otherwise, these statistics are just a giant status quo amplifier.
Re: Attacking discrimination with smarter machine learning
#9But how can we find out if a company uses ML in a non-discriminating way? If we cannot see it, and check it, then there is no incentive for companies to use it. My guess is that at most companies spending time on making an algorithm non-discriminating will be viewed as a waste of time and money.
I think companies like Google need a team akin to NYT's public editor. Obviously not perfect, but better than nothing.
Re: Attacking discrimination with smarter machine learning
#10But how can we find out if a company uses ML in a non-discriminating way? If we cannot see it, and check it, then there is no incentive for companies to use it. My guess is that at most companies spending time on making an algorithm non-discriminating will be viewed as a waste of time and money.
> But how can we find out if a company uses ML in a non-discriminating way?
It is an interesting question. However, there are ways to answer this question as we have been measuring discrimination before algorithms. Algorithms are a way of reaching a decisions. Therefore, we can measure discrimination if we audit decisions generated by algorithms. In other words, look at the input-output of the algorithm and measure the impact. In the US, the doctrine of disparate impact has long been used a guideline to evaluate discrimination [0].
[0] https://en.wikipedia.org/wiki/Disparate_impact#The_80.25_rul...