Rise of the machines: is artificial intelligence about to have its day?

My favourite technology story from the past week has to be the epic battle between Google’s AlphaGo and Go grandmaster Lee Sedol, which ended in a decisive 4-1 victory – in the computer’s favour – on 15th March in Seoul.

 If you missed it, the short summary is that the Google DeepMind collaboration built AlphaGo with the aim of mastering the ancient Chinese game Go. It relies on intuition and has been widely viewed as a ‘grand challenge’ for artificial intelligence. AlphaGo is programmed to mimic expert players of the game, and learn from games played against itself. The first strong public indication that it was succeeding came last October, when it won a 5-0 series against three-times European Go champion Fan Hui.

 I couldn’t wait to see what happened this time and really found it hard to decide who I was rooting for. On the one hand, it is just incredibly exciting to see how advanced AI technology is becoming. Who would have thought that a machine could really learn to act in what we perceive to be an intuitive way? It’s something we associate with living beings after all.

 And yet, how could I fail to sympathise with Lee Sedol’s frustration and incredulity at becoming the underdog of the match? This is something that was predicted to be ten years away, and, in fact, only a few years ago would have been inconceivable. What a dilemma!

 The dilemmas don’t end there, by any means. From the earliest days of our imaginings about intelligent machines, the question of ethics has been a central one. Who could, or should, ultimately be in control? Commenting on the Seoul competition, an opinion piece in the UK’s Guardian newspaper wondered: “as AlphaGo and AIs like it become more sophisticated – commonly outperforming us at tasks once thought to be uniquely human – will we feel pressured to relinquish control to the machines?”

 It’s a good question. We have already incorporated millions of automated processes – which we use more or less unthinkingly – into our everyday lives. Just look around your home. At your fridge, your coffee maker, your central heating system. And as developments continue, such as those in the military like the so-called ‘Robo Wars’, which will see the Ministry of Defence host a ‘large-scale demonstration of maritime autonomous systems’ (that’s drones and seacraft to you and me) this autumn, it becomes more and more pressing.

 Even a seemingly straightforward application of AI can obscure the lines of responsibility. Who is at fault if a drone strike goes wrong or a driverless car has an accident? A paper by M. Mitchell Waldron pondered this ‘question of responsibility’, and the same issue was laboured over by none other than Isaac Asimov, the prolific author and creator of the ‘Three Laws of Robotics’, which were intended as a framework to underpin the behaviour of robots designed to have a degree of autonomy.

 I think we are some way off horrors such as those described in the classic sci-fi story ‘With Folded Hands’, where robots ultimately leave humans with little to contribute to the world. Or indeed the adverse, as seen in Spielberg’s 2001 film A.I., where a human child acts in the traditional ‘baddie’ role, duping a childlike android into an act designed to cause it problems.

 I’m choosing to stay optimistic about a future where AI plays a legitimate and useful role. My main interests for now are two-fold:

1. How to create the type of 5G-enabled, interactive environments that I spoke about a couple of weeks ago, where the machines in the ecosystem have the capability to make intelligent, intuitive decisions and carry out actions based on those decisions, with lightning-fast reaction times.
2. How we can make AI work to protect customer interests in the commercial world, by tackling fraud through monitoring credit card systems, or protecting personal data by detecting cyber security threats for example – both areas where the telecoms industry has found weaknesses. There are plenty of opportunities to be explored.

I think we can find a way to make intelligent machines a functioning and functional part of human society, in an ethical way. Maybe it all comes down to intent. DeepMind’s strapline is:  ‘Solve intelligence. Use it to make the world a better place.’ That seems a good starting point.

 What are your views?

Photo credit: A Health Blog