Will 'AI do a better job than anybody?'

During our interview with Lord Neuberger on his patronage with the Aberdeen Law Project, he kindly shared his thoughts on the future of law with regards to AI.   

"I am regretfully of the view that in due course, one day, AI will be able to be programmed to do a better job than anybody doing anything. I may be wrong, and I am not speaking as anybody with any great authority, but it seems to me that all we’ve got in our brains is chemistry and physics.

Artificial intelligence works in a different way – it’s all wires and electronics. Our brains aren’t electronic: they are electric and there is chemistry going on there, but there is no electronics. But in the end, computers mimic us and, in the long run, I suspect – whether it's 10 years or 200 hundreds – they will be able to make a machine that mimics the human brain, including emotions, but which can be adjusted so that the emotions are suitably calibrated.

That’s guesswork. More to the point, I think it would be useful in due to course to have – whether it’s AI or some other form of computerised system – a system which a judge can plug in the various features of the case and get an idea of what the sentence should be. He or she might then say, well I’m not very happy with the results that provides, but at least it will give some sort of baseline approach and would make the judge think: why am I departing from that? If the machine says eight years and the judge is thinking of 15 or 2 years, it will make the judge think, am I going wrong somewhere? I think to that extent, it is a useful check. Whether we go on to the next stage is another matter. But I would have thought it was useful to do that.

They have already done that in some American states to an extreme, in the sense that they have a grid and you just put in the various features of the case and out comes the answer. Whether that approach is still pursued, I don’t know, but I think that is very dangerous. I think the human element has its problems – for example, if a judge is emotionally affected by something like having a quarrel with a loved one that day, and that then has an effect on the judge’s decisions. But on the other hand it does allow you  to adjust in an emotional way that is appropriate.

That’s why I quite like the idea, when machines are really up to it, to use machines as a guide, as a check. So, in effect, the judge can say to him or herself, I am departing from what the machine says for reasons a, b and c. And if reasons a, b and c, when you put them down look a load of rubbish, you can think, maybe I shouldn’t be departing from the machine. It’s why it’s so important in my view that judges should always give reasons for their decisions, because it is only when you start looking at your reasons that you start wondering whether you are right and the people can see why. The public is entitled to know why a rapist is not sent to prison, and a rapist is entitled to know why he is sentenced to 20 years rather than 10. It’s important for the judge to articulate to him or herself, and to the public, why the sentence is what it is."

Now it's over to you, we want to hear your thoughts on the future of law and AI.

How much do you see AI playing in the future of law and the legal profession?