An invitation for co-development or I go it alone…
[This also being a copyright / idea claim]
- Undecided what name will stick; either
Ethics Test Services, or
Autonomous Judgement/Decision Analysis Services;
- Because it is about checking the morality baked into, or emerging from, algorithmic decisions and/or decisions and conclusions from autonomous and self-learning systems;
- Contra “Computer says No”, obviously.
If you’d want to learn what that refers to; see here;
- [Intermission] Whereas some in European politics (sic) discuss to impose a limit where autonomous systems without one human in the loop anywhere would have to have an ‘explanatory’ function that can display in layman’s terms how it arrived at some decision, and that being contestable. But the questions are: What if the ‘system’ were hosted outside the EU (and just like inflation, Gresham will obviously apply), and what if (maybe ‘when’; we’re talking politicians here) such a very first step towards transparency may still not make it, and what if as a cheap escape trick the human would and could only click ‘OK’ — could (s)he be culpable?
- Elements would be:
- Process correctness,
- Data correctness,
- Exceptions handling; essential and necessary.
- This, in Standard Form and with an overall human (me; run to the hills) judgement both over process/systems quality and over moral/ethical admissability;
- Will have to extend the notions of ethics, morality et al. here; e.g., how humans make decisions in the first place with all their errors of all kinds, what to do when systems/humans don’t follow morality and/or the decisions from the systems.
So, everyone (dabbling in this space from now on,) will pay me serious license fees for using the above ideas in commercial services… [note: I’m serious]
And/or all help is welcomed.