Force 9 gale, training and jet lag – but still thinking about Machine Learning and AI

Some work-related things

The last few weeks have been a mix of upgrade projects, professional development, a force 9 gale, 8-hour sea journey to Ireland and turbulent air travel from America, with a pinch of jet lag. Hence the slightly reduced output as far as this blog is concerned.

This is a bit of a teaser post, just to let you know that I am working on presenting – at some point –  some advances in machine learning and AI that my company are working on – but first, I need to actually do the work so that I can report an authentic experience.

I’ll simply hint that I have heard of a ML use case at scale with promising results, using some pretty powerful methods, and that I know of another use case that could possibly be as interesting.

I’ll repeat what I said in a previous post: I feel that with the maturity of data provision (good warehouses, faster methods, better understanding, better tools) we are now in a position to deploy new and interesting software to add predictive and deep learning capabilities to our existing BI systems. The move from forensic reporting to AI-based forecasting, scoring and automation is afoot, and I hope to report back on this more detail in Q1 2019.

More thoughts about Cynthia

Enough about work – I have also been working on my Cynthia concept, which is a set of models that will interlock to form a synthetic personality. I made a post about Cynthia (article on LinkedIN – Introducing Cynthia) where I describe a basic personality framework.

I have also made an attempt at modelling an influencer mechanism, which would use strategy to bring about an outcome. It may be an application that interfaces between Cynthia and a human, or it may evolve into a survival mechanism for an AI such as Cynthia to overcome obstacles and achieve goals.

I show the high level concepts for this influencer mechanism which I have called Machiavelli, for obvious reasons.

Worth labouring the point: The interesting thing about the model above is that it can be instantiated in different ways: As a human-based system, as a hybrid of humans with application assistance, as a stand-alone goal seeking system or as a component of an aggregation of such modules which would eventually become Cynthia. What will happen will heavily depend on time and possibly funding.

That’s it for now – it’s going to continue being busy so I may not be back for a few weeks – but I’ll try and make it  interesting when I do return.