AI predicting ‘criminals’ based on their face

unkategorisch

Harrisburg University currently incinerates this paper (PR for it via archive.org). I wonder how people can spend months working on a concept like this and don’t prepare for all the not so pleasant attention it will surefire gather upon its surfacing.

I have no idea about the content. If something does not get published it can not be examined.

I would not be surprised if there would be two sides looking at this: One says that it is not morally right to conduct such an investigation. The other will say that it is not morally right if certain thoughts and research directions are not allowed. Discussions of this nature lately not often end up in an amicable exchange of each others viewpoints.

Not related to this I personal find trouble in the term “criminals”. “people exposing criminal behavior” I like more. The later word is a determination. A person is so many cm tall, has a certain eye color and is – well – criminal or not. Which is wrong.

Of course most criminal behavior emerges from the same perpetrators. It should always be the criminal action that gets punished, not the person per se. A society where the government instruments strong deterrents against such behavior benefits in very great measure. Even better: All people – regardless of their wealth – will have a better life. It would be stupid and counter productive ignoring the reality of said behavior clusters around certain people. But it should always be the actions of a person that get punished. Not more, not less. Actions happen in the real world, they can weight and judged by others, and therefor rightfully punished. I personally like the sound of the German ahnden actually better than that of its English equivalent.

These things are known since 200 years or more. Sometimes I wonder for how much longer. If we cancel research – only because the initial premise is stupid and not fitting dogma – we might hasten the demise of knowledge.