A retrospective on Artificial Intelligence Chapter 2

After several years during which I was able to dedicate myself to Artificial Intelligence, actually to modelling and development of neutral networks, only during the free time, I managed to be back on the subject only 2015. A main topic was to understand the difference between Artificial Intelligence and Machine Learning. 

Before that time, I always spoke about the aforementioned neural networks, sometimes also of reinforcement learning, but machine learning was a buzzword I previously seldom used. Although I never ceased to read the most recent scientific papers as well as the fundamental texts of the AI pioneers. Was that just marketing? Actually it is not, because the first paper mentioning Machine learning dates back to 1949. Donald Hebb, a Canadian psychologist who focused his studies on the function of neurons, used the concept of machine learning in The Organisation of Behaviour: A Neuropsychological Theory, a book that is available online and is a recommended reading for a deep dive. 

An important point is that I focused much more on a specific application of a specific branch of Artificial Intelligence, Bayesian Neural Networks. On the contrary we can consider nowadays Artificial Intelligence as a science with specific theoretical constructs and different branches of applications. 

The most renown of which are Machine Learning itself, Natural Language Processing, with the chatterbox being one of the most useful products, Pattern Recognition, that is the base of face, speech and handwriting recognition. Robotics and automation constitute, I am really firm about the matter, also a component of Artificial Intelligence, involving adaptive systems able to interact with the external environment, more or less dynamic and subject to changes and events. 

Such a discussion was however revealing how the business seriously started to nourish interest for AI and moved me to intensify my studies, that I always conducted in parallel with my profession, in order to be ready once the market would have matured enough. 

Indeed RPA, or Robotic Process Automation, began to be experimented and put in place by main enterprises, a revolution that would have impacted the internal workflows of a company. Indeed the first attempts were oriented to disrupt as less as possible of what already existing. The RPA tools should have learnt from the current operators with the goal to replicate their actions, for instance the routine tasks that a controller or a customer support manager executes on the screen while working with popular solutions such as Excel. The most repetitive tasks should have been replaced soon by RPA bots, reminding me the time I wasted with the macros of Windows 95 (and the previous versions) while at the high school. 

Apart the results in terms of efficacy and efficiency, the biggest impact of RPA is that business analyst and consultants must start to take notice of automation, not just by software and system engineers. That has been the most efficient way to make the management aware of the potential of AI, or, better to say, the applied branches of AI. 

Meanwhile the trivial ethical problems related to autonomous driving started to find a place on the headlines of popular newspapers. I call them trivial because ethics and AI are not limited to the decision whom to keep safe in case of a car accident. It is a much complex discussion that involves the epistemological aspects of AI itself. 

The main universities all around the world are currently coping with the topic. In Frankfurt am Main, the town where I currently live, the leading mind behind the Ethics and AI program at the Goethe Universit?t is prof. Zicari. He is also Italian and previously worked at my Alma Mater, the Politecnico di Milano. The subject has such a relevance that the online content of the course has been made available also outside the official walls of the well renown academical institution. With this I consider completed the introduction to my articles, starting next week I will analyse the current stand of AI. 

#artificialintelligence #rpa #machinelearning #aiandethics #donaldhebb #theorganisationofbehaviour #neuralnetworks #nlp #pattern recongnition #roboticsandautomation