“Artificial intelligence” is everywhere these days; it’s the world’s most popular oxymoron. Let’s break it down into its constituent parts. Artificial intelligence is certainly artificial. But is it “intelligence”? I am not so sure.
AI, we have been led to believe, is going to take away all of our jobs. Is it going to take away mine? Eventually, perhaps, but I suspect not quite yet.
(I have heard that someone recently asked a judge if I would lose my job to AI, to which the judge replied, “Sixty percent of the legal profession will probably lose their jobs, but Stan will be fine”. Good to know.)
I was working on a judgment recently in which I found a sentence that had the following structure:
“Where Legislative Provision A or Legislative Provision B applies, Consequence C is the result.”
On my final read through the judgment, it dawned on me that the use of “where” caused the sentence to be not quite telling the story the judge was wanting to tell.
Legislative Provisions A and B made up the entire universe of possibilities in the field of discourse the judge was operating in. (In other words, in any given situation one or other of those provisions is going to apply.) The reader would know this. Thus, to say that C is the result of A or B applying wasn’t saying very much at all; C is going to happen in any event. In a world where one or other of those provisions is always going to apply, what does it add to say that that Consequence C is going to happen “where one of the two provisions applies”?
What the judge was really wanting to say – or so it seemed to me – was that it doesn’t matter which of Legislative Provision A or B applies; Consequence C is going to happen. The judge, presumably, was making a point. Judges don’t usually say things just for the sake of saying them. The judge, I inferred, was making a positive statement that they aren’t worried about which of the two legislative provisions apply. It might have been that the judge was purposely not engaging with the question of which of them applies (for example, to dodge an argument as to which of them does apply) or was just doing that thing judges do of not engaging with something they don’t have to engage with (or would rather not). There is a hidden “I don’t care” buried in the sentence, which was not being conveyed by the word “where”.
In other words, it was a “Whether” sentence, not a “Where” sentence.
Grammatically and structurally, the sentence worked perfectly well as written. It was only by thinking to myself, “what exactly is the judge saying here?”, that I was able to get to the point where I could forensically attack the word “where” and form the view that it was the wrong word. Organic Stan was picking up a vibe. Would AI Stan have been able to do that? Does AI do “vibes”?
Anyway, I ran my argument, and “where” was changed to “whether”.
This is why I am not (yet) too scared of AI. It deals in facts – quite possibly more thoroughly, and better, than a human could – and I can imagine, indeed I might even welcome, a world in which I could give over to it the half of my job that involves checking facts, quotes and references. Have at it, Mr Machine Learning.
But I’m not sure it is equipped to deal with nuance – to intuit that the “where” in that sentence should have been “whether”.
Full disclosure: I haven’t yet seen for myself what AI is actually capable of. I might be unpleasantly surprised. Still, it seems to me – vibes, again – that, at least as we sit here now, as a tool for writers (or judges!) AI looks like a somewhat blunt instrument.
Of course, the future is unknowable, and things are capable of developing rapidly and in unexpected ways. (“Ya don’t say.”) It may well be that in five years time AI will come for me, and I will find myself lining up for a place on the starliner Axiom, to join all of the other redundant humans sitting around in air conditioned comfort, sipping on an ersatz mai tai and gazing down wistfully at the planet that we have left in the care of our cyber overlords.
What a time to be alive.

Leave a comment