[liberationtech] The Invention of "Ethical AI"

[Andrés Leopoldo Pacheco Sanfuentes]

The concept of “risk” may be deemed VERY discriminatory. Minimizing “risk” discriminated against populations considered “risky” when the “risk” is unavoidable for that population. Like poverty.

In Asylum law there’s this concept of “Particular Social Group:” << It is also often described as a group sharing a common characteristic that is so fundamental to their individual identities that the members cannot—or should not be expected to—change it.>>

And then of course the suspect circular references to “recognized” or “neutral” third parties...

Regards / Saludos / Grato

Andrés Leopoldo Pacheco Sanfuentes

> On Dec 25, 2019, at 2:26 PM, L Jean Camp <ljeanc at gmail.com> wrote:
> 
> 
> >An old saying: "When you hear the word 'ethics,' grab your wallet."
> Among people who think their wallets matter and other people's lives don't. Merry Children in Cages Christmas.
> 
> Anyway, to the actual discussion, the link to the proposed rule is:
> https://www.consumerfinancemonitor.com/wp-content/uploads/sites/14/2019/08/HUD-Proposed-Disparate-Impact-Rule.pdf
> 
> And I think a very good description is here:
> https://www.consumerfinancemonitor.com/wp-content/uploads/sites/14/2019/08/HUD-Proposed-Disparate-Impact-Rule.pdf
> 
> For those who do not like to download PDFs here is the AI component. Basically building a discriminatory model is only subject to sanctions if the claimant can prove that the model was built with intent to discriminant:
> 
> 6. Where the cause of a discriminatory effect is alleged to be a model used by the defendant such as a risk assessment algorithm, the defendant can:
> I. Identify the material factors used in the model and shows that these factors are not substitutes for protected  characteristics and that the model is predictive of credit risk or another similar valid objective
> II. Show that a recognized third party that determines industry standards is responsible for producing, maintaining, or distributing the model, the inputs within the model are not determined by the defendant, and the defendant is using the model as directed by the third party
> III. Show that the model has been validated by a neutral third party that found it to be empirically derived, that its inputs are not substitutes for protected characteristics, and that it is demonstrably and statistically sound and accurately predicts risk or other valid objectives.
> 
> 
> 
> 
>> On Tue, Dec 24, 2019 at 1:18 PM John Young <jya at pipeline.com> wrote:
>> An old saying: "When you hear the word 'ethics,' grab your wallet."
>> 
>> 
>> 
>> -- 
>> Liberationtech is public & archives are searchable from any major commercial search engine. Violations of list guidelines will get you moderated: https://lists.ghserv.net/mailman/listinfo/lt. Unsubscribe, change to digest mode, or change password by emailing lt-owner at lists.liberationtech.org.
> 
> 
> -- 
> Prof. L. Jean Camp
> http://www.ljean.com
> 
> 
> Research Gate: https://www.researchgate.net/profile/L_Camp
> DBLP: http://dblp.uni-trier.de/pers/hd/c/Camp:L=_Jean
> SSRN: https://papers.ssrn.com/sol3/cf_dev/AbsByAuth.cfm?per_id=262477
> Scholar: https://scholar.google.com/citations?user=wJPGa2IAAAAJ
> Make a Difference 
> http://www.ieeeusa.org/policy/govfel/congfel.asp
> -- 
> Liberationtech is public & archives are searchable from any major commercial search engine. Violations of list guidelines will get you moderated: https://lists.ghserv.net/mailman/listinfo/lt. Unsubscribe, change to digest mode, or change password by emailing lt-owner at lists.liberationtech.org.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ghserv.net/pipermail/lt/attachments/20191225/34daca16/attachment.html>