Ethics in AI – two short notes

We are introducing AI in our search systems at the library. Nothing new if you ask me, we have had it for years, we just called it machine learning.

That leads to colleagues advocating for us looking into the ethics of AI. Again – that should not be something new. And frankly I am not quite sure that it is that relevant in a library setting. At least as we are actually going to use AI.

Anyway, two short notes – so I have them somewhere and, wont forget them:

  1. The ML algorithms are working on the statistics we feed them. Not the reality we would like to have. That is note: Both the difference between the statistics we feed to the algorithms and reality. And the difference between the reality we actually lives in, and the reality we would like to live in.
  2. The algorithms reflects (with the filter between reality and statistics mentioned above) the reality of yesterday and, with some probability, the reality of today.