181–190 of 222 posts
Six percent annual turnover for non-compliance seems too low. Should be six percent for first offense, 12% for second, 25% for third, etc. Until the company fixes it's compliance or becomes insolvent.
This opinion is terrifying to me. Why has our culture become so authoritarian?
Earlier quoted context omitted.
We're doing credit scoring at the Danish bank i work at. One of our requirements is that the model and architecture has to be able to provide explanations for why yohr rating is whatever it is. Both to regulators, internal auditors, and customers. Personally, i thimk denying people a loan is a pretty impactful decision on peoples life. They deserve a reason.
Well, 90% yes and 10% no, in my opinion. Yes, because of the obvious ramifications, as you've alluded to. No, because it's a private company you're asking the loan of. If they don't want to lend to you, then they don't have to lend to you. I believe they should be able to come to this conclusion however they like, except based on the obvious factors like culture, gender (or lack/fluidity of), etc. but mostly yes. I t…
Really? How do you figure? I genuinely cannot tell how you are reconciling these two points of view. While should a private business be able to do business however it wants, but somehow be restricted when it comes to this list of 3-4 factors? Can you elaborate on that?
More specifically, how do you reconcile this opinion with the fact that many algorithmic models do perpetuate existing discriminations against these groups that you listed, despite these things supposedly not being parameters of the algorithms?
(Note: I am not condoning discrimination. In fact, I personally don't think that private business should be able to do whatever they want, and that they should be much more restricted than they are today in many, many different respects.)
Earlier quoted context omitted.
Rules around fake information scare me because they're a limit on speech, and as Russia has recently shown, fake information is anything a dictator doesn't approve of.
Russia has recently shown, fake information is anything a dictator doesn't approve of. This isn't an issue of "limits on speech", but rather, another reminder that one shouldn't enable folks to become dictators. Not having some reasonable limits on actual misinformation makes us all less free, however, because we cannot put our trust in some organizations.
If you read the article they're asking for more than explaining algorithms. Overall they want the the tech providers to be responsible. Explaining algorithms could, in theory, give away a competitive advantage. However fairness to users seems to be a priority in this decision.
>> "Large online platforms like Facebook will have to make the working of their recommender algorithms (e.g. used for sorting content on the News Feed or suggesting TV shows on Netflix) transparent to users. Users should also be offered a recommender system “not based on profiling.”" Both of those seem like good ideas and progress. The non-profiled recommender system option especially! It's also really bothered me th…
Care to elaborate? Discrimination in terms of what ads are displayed perhaps?
If you read the article they're asking for more than explaining algorithms. Overall they want the the tech providers to be responsible. Explaining algorithms could, in theory, give away a competitive advantage. However fairness to users seems to be a priority in this decision.
I would love to first see a technical definition of fairness from EU that can be used to evaluate algorithms. That is a non-trivial detail often overlooked from these discussions.
> Article 29 Recommender systems
> 1. Very large online platforms that use recommender systems shall set out in their terms and conditions, in a clear, accessible and easily comprehensible manner, the main parameters used in their recommender systems, as well as any options for the recipients of the service to modify or influence those main parameters that they may have made available, including at least one option which is not based on profiling, within the meaning of Article 4 (4) of Regulation (EU) 2016/679.
> 2. Where several options are available pursuant to paragraph 1, very large online platforms shall provide an easily accessible functionality on their online interface allowing the recipient of the service to select and to modify at any time their preferred option for each of the recommender systems that determines the relative order of information presented to them.
Earlier quoted context omitted.
Rules around fake information scare me because they're a limit on speech, and as Russia has recently shown, fake information is anything a dictator doesn't approve of.
Russia has recently shown, fake information is anything a dictator doesn't approve of. This isn't an issue of "limits on speech", but rather, another reminder that one shouldn't enable folks to become dictators. Not having some reasonable limits on actual misinformation makes us all less free, however, because we cannot put our trust in some organizations.
Six percent annual turnover for non-compliance seems too low. Should be six percent for first offense, 12% for second, 25% for third, etc. Until the company fixes it's compliance or becomes insolvent.
This opinion is terrifying to me. Why has our culture become so authoritarian?
I see a vast technical writing documentation project in their future.
But they must be short and easy to understand by users. Like this: "Our algorithms use gradient descent. Data flows through our connected tubes, slowly wiggling their size until the data starts flows back and forth faster."
That's going to be interesting to see if the whole credit scoring industry will have to disclose their algorithms as well. Looking at you, Schufa.
My understanding regarding credit score is that it's one of the most regulated and explainable algorithms. It's not transparent to the public, but it is auditable due to anti discrimination regulations. Am I wrong on this?
You have a right as a consumer to the underlying data from the credit reporting bureaus, but not the proprietary algorithms that determine risk.
I understand the desire for transparency, but at its core credit scoring is fraud prevention. It is like asking Visa to explain what criteria they use to determine if a charge is fraud, which predominately helps the people trying to do the frauds.
That's going to be interesting to see if the whole credit scoring industry will have to disclose their algorithms as well. Looking at you, Schufa.
My understanding regarding credit score is that it's one of the most regulated and explainable algorithms. It's not transparent to the public, but it is auditable due to anti discrimination regulations. Am I wrong on this?