Engineers all over the world are developing robots that are tasked with a single horrifying purpose — to kill people.
There are flying robots, robots that hurtle like rockets through the sky until they find their target, robots that patrol borders, surveillance robots, robots that co-operate with other robots, and robots that identify human targets and summarily destroy them. Purely as engineering achievements they are hugely impressive. As examples of man’s indifference to the ‘do no harm’ principle, they are sickening.
Let’s be clear: if you are ever on a weapon’s hit list your days are numbered. It will not give up until it has found and destroyed you. There will be precisely nothing you can do about it. And ultimately we could all be on a list somewhere.
What makes these weapons so much more deadly than ‘traditional’ weapons is that they are, and are intended to be, both autonomous and intelligent. But their intelligence is not like ours. They never get bored or tired, they never give up, they have neither fear nor compassion, and they make their own decisions.
Autonomous robots are unpredictable by design. This is the difference between artificial intelligence and mere algorithms. They are not simply following rules, like an airliner’s autopilot. They are given a task to do and work out for themselves how to do it. And the processes by which they do this are neither understood nor understandable. Algorithms are sets of instructions which, however complex, can be analysed and understood, but artificial intelligence is embodied in highly complicated self-organising electronic networks, just as human intelligence is embodied in the human brain. AI can be trained, but just like trained animals or even humans, training is never precise; the learner has to translate the training into action. Faced with a new situation, one that was not included in the training perhaps, or one that was included but was misinterpreted, there is no way of knowing what will happen.
And of course they can simply malfunction, like the Boeing 737s that insisted on diving into the sea regardless of what their pilots directed them to do.
Sadly none of this is science fiction. At least sixty-four countries are already developing or operating ‘unmanned aerial vehicles’ (UAVs), and although some UAVs are currently controlled (albeit remotely) by humans, artificial intelligence (AI) is increasingly being used to remove the need for human supervision altogether.
There has to be a push-back against this utter absurdity. Ultimately we need to change the world. Of course we do — as long as the dominant forces in politics are the politically forceful and as long as we continue to see each other as potential enemies rather than potential friends, there will be preparations for war, and these will spill over from time to time into actual conflict.
The Peace Party patiently does its bit to help bring about such a change, in the knowledge that others are working to the same end and in the hope that eventually the change will come. But we can’t afford to await that change. Autonomous weapons are in the sky, on the ground and in the sea now.
Arms manufacturers live in a different world, strangely insulated, it seems, from ethical, humanitarian and even rational considerations. They have a lot of money, a lot of power, and a lot of influence. A peaceful world would not be in their interests. As long as countries will buy and deploy their weapons they will continue to develop and build them.
But there is a push-back. It comes partly in the form of the redoubtable Mary Wareham and Jody Williams who were awarded a Nobel prize for their work in bringing about the banning of landmines. In their words though this was ‘chump change’ compared to stopping the development and deployment of killer robots.
They now work with the Campaign to Stop Killer Robots. More than 100 organisations in 54 countries have joined the coalition and the aim is to get an international treaty signed by 2021. It won’t be easy, but people said banning landmines would be impossible, and it happened.
As Jody says,