Re: AI KR Role 2. identify and prevent Bias

Dear Paola,

I applaud your emphasis on trying to eliminate bias through appropriate KR methodologies. Unfortunately it cannot be eliminated completely. Bias can also be caused by factors inherent to the observer in the process of applying the scientific method.
Formal logic and any formal computational framework used for AI and KR will run into problems if they go beyond dealing with narrowly defined domains of discourse with corresponding ontologies, KR and algorithms, knowledge graphs and object based formalization.

In our modern daily life most human activities can be modeled as complex adaptive systems or sets of complex adaptive systems, where interaction processes come to the fore as the principal ingredients for modeling, and to a lesser extent clearly defined objects.
Just as a thought experiment, try to imagine setting up two formal frameworks for (1) general relativity,  (2) quantum physics and the corresponding AI and KR formalization for an AI to reconcile scientific method, observations, experimental data, inference engine, and theorem proof systems.
Not possible, bias can be eliminated in both separate systems, both trying to combine them introduces a bias.
As I would like to point out, only certain types of knowledge can be made bias free, and consequently using KR have functioning AI for these.
Unfortunately in our modern complex technological world, the knowledge so often required must take into consideration (sets of) complex adaptive systems or knowledge types that do not easily lend themselves to unbiased KR.
Eliminating bias is a worthy endeavor, but it will invariably lead to a limited subset of possible knowledge domains that can be formalized through KR to yield AI that deals with as little bias or no bias at all.
If you are OK with that, fine. But as a mathematician I am not. And I am sure most advanced brain research is looking for avenues to make the modeling of the "hard" knowledge domains possible, and to have ways to deal with bias in a satisfactory way.
regards


Milton Ponson
GSM: +297 747 8280
PO Box 1154, Oranjestad
Aruba, Dutch Caribbean
Project Paradigm: Bringing the ICT tools for sustainable development to all stakeholders worldwide through collaborative research on applied mathematics, advanced modeling, software and standards development 

    On Friday, June 12, 2020, 2:08:52 AM ADT, Paola Di Maio <paola.dimaio@gmail.com> wrote:  
 
 (role 1 is to identify and naming the classes, entities etc) as per previous emails

Role 2;  To identify and prevent representational and algorithmic bias - https://docs.google.com/presentation/d/1zYhjFQ3CYdGZDZgvgxnhCguTomyJuD-lpNgv1pzo5yI/edit?usp=sharing