Artificial Intelligence – A primer on AI : Tony Bloomberg

A simple definition of AI is ‘the digital simulation of human intelligence’. AI is said to be the 4th Industrial Revolution (4IR), after steam engine, mass production and digital technology. The media is meeting this moment with something between hype and hysteria. I will not attempt to explain the different technical flavours of AI. Rather I will try to draw a picture of the AI ecosphere, as I understand it, in the hope that it will be useful.

Applications of AI can be seen already in different spheres, and these will increase and deepen, including: Industry eg robotics, Transport eg autonomous vehicles, Marketing eg sales targeting, and Finance eg algorithmic trading. Other applications, of more UNICEF interest, include: Health eg remote diagnosis and prescribing, and Education eg remote personalized learning. AI depends on learning from ‘Big Data’ supplied by users, including of Google and Facebook; these volumes are increasing exponentially in all parts of the world.

The technologies supporting AI continue to be developing rapidly. Computers have high penetration, mainly only in industrialised countries, but data usage has moved rapidly to smartphones. Even in Africa, mobiles now cover about half the population. Mobile 4G has become the telecom standard in industrialised countries and 5G, offering faster broadband is set to supersede that. Presumably that trend will follow in developing countries. So far, big data centers offering cloud computing are only found in industrialised countries but these providers are already planning to decentralize to Fog (nodes) and ultimately Edge (device) networked capabilities. The Internet of Things (IoT), where devices are embedded in retail equipment from lights to washing machines, is still in its infancy but set to increase rapidly.

Regulators have finally woken up to this AI 4IR, including recognition of ethical issues involved. Quickest off the mark are the EU that has an ‘EU AI Act’ drafted that must now be discussed with its members. The USA White House has drafted an ‘AI Bill of Rights’ and there is some draft legislation waiting for Congress to get serious about considering serious things – not likely in the short term I believe. Meanwhile global competition is pushing the technology way ahead of regulation, and some countries (eg China) will see regulation differently from others. Legislators, at least in USA, are scientifically illiterate and my main hope is that technology companies will be forced to adopt EU standards.

Social adoption of AI will be slower than technology companies would like, including in industrialised countries with ageing populations. But judging from my oldest grandchildren that will not be the case for the younger generation. Legacy systems and infrastructure will also be a drag on AI introduction, with those countries investing more in the public good taking a lead – think smart roads and AI power grids. There will of course be disparities between population groups and countries – we have seen plenty of examples of that in IR1, IR2 and IR3 so why not IR4. Still, from my retired armchair, I am looking forward to seeing what transpires.