This folder contains some thoughts and analyses which try to make sense of the claims and some of the counter-claims being made about AI.

I recall one of my first projects at CAP Scientific when I was discussing with Mike Christie how our system for Sensor Data Fusion (the Alternative Tracks Method) using Bayesian combination was better than the alternative Blackboard AI system (developed by Nigel Shadbolt then at Nottingham).

In chatting to Mike about why our system was better than the Blackboard (and it was!), it was only when I realised ‘why’ (the Bayesian formulation acted as an adaptive gain control) that everything made sense. This led me to a life long principle - that to fully understand abstract and mathematical representations of anything you have to recognise the fundamental physical/material mechanism that the abstract concepts rely on, and through which they interact with the real world.

Anything else is a hallucination.

The pieces here are my initial thoughts and exercises which, if making sense, will form the basis of a paper/presentations submitted to the next OR Society conference.