AI is here – Should we be very afraid?

The expression “Artificial Intelligence” (AI) is being mis-used. Many recent articles on AI have launched straight into a discussion about robotics, which is merely one application of AI, incorporating a machine to perform physical jobs (e.g. welding a section of car to the chassis), rather than a purely intellectual one (such as assessing a mortgage application).

Robotics discussions about job losses get people heated by allowing them to visualise a robot sitting at their desk – this is b*ll*cks (sorry but it raises passion in me too). Journos and Marketeers love to get us all razzed up about being replaced. It is not a revolution, as the scaremongers would have it, it’s evolution. We will not awake one Monday to find our office full of little R2D2s and C3POs. This is not what the brave new world looks like – the office, and the paperwork, disappears altogether.

AI, the phrase was coined by John McCarthy, a US computer scientist in 1956 (61 years ago!). There are two types of AI – weak and strong.  A “Weak” AI system is capable of responding to a set of pre-programmed variables and respond in a specific way (like workflow). When people talk about AI in 2017 they generally mean “Strong” AI systems, which can respond to a new set of variables and decide on how to respond. A good example here would be driverless cars, where it is impossible to predict every set of circumstances beforehand.

AI systems can respond to a “new” set of variables (e.g. identifying a contextual pattern in documents) in one of two ways – it can either suggest appropriate action, requiring human intervention, or can learn directly and react appropriately, storing this experience for next time. Robotics which use Strong AI are perhaps the most interesting (and frightening) area of development, as it gives rise to the scenario where a physical machine is making a judgement call, based on a new set of circumstances. Can we be sure what the reaction will be? Big Data allows us to trial many more potential situations but is not fool-proof. What will your driverless car do when faced with an angry moose, attacking vehicles on the motorway?

Progress is evolutionary, and AI has been replacing people for 50 years. There are many jobs that used to be done by people, which are now done by machine – probably the most startling example of this is shopping. AI has allowed online shops to suggest items to buy based on your previous behaviour, as well as supporting the sale, and delivery of goods.

Many commentators say it is not yet time to panic. I am not sure what good panicking will do, but the “evolution” has accelerated (look at how quickly the high street has been destroyed). To date AI (and computing in general) has created more jobs than it has removed, but there is no guarantee that this will continue to be the case.

The power of AI is in the combination of technologies it unleashes – big data, voice recognition, natural language and facial recognition. The potential is huge, and progress will be driven by competition in various markets. The frequency of new start ups means established business must change or die. All good stuff? AI will certainly contribute to growing the global economy. The two elephants in the room though are security (easier to hack machines than people) and how we, the human race, adapt to fill our new niche, and continue to add value.