Why AI will never replace a good leader: or employment lawyers.

Sorry, machines: we’re just better than you

My friend and Partner Jocelyn Paulley sent me this piece, which suggests that one of our Titans of Tech believes he may replace himself as a leader in one-to-ones by an AI-powered Avatar. It is said this is to avoid the bits of the job he doesn’t like: people.

There’ll be more to it than that, of course there will, but I wonder if the central conceit is in any event fundamentally misguided.

I don’t say this from my passionate belief that to be a person is not something that can reduced to 0’s and 1’s. It is not because I think – though I do think – that to be human is to be touched by the Divine, whatever you perceive as the Divine, supernatural or otherwise: ‘otherwise’, in my case, but this is a legal blog not a religio-philosophical one. Rather, I appeal to the God of AI: that is, data.

It all comes back to data and the processing of data, whichever model of AI is (forgive me) prayed in aid. The machine takes the numbers and crunches them, that’s all. There is no magic.

So what then is the data when it comes not to people, but to a person? ‘People’ are much easier. Machines can predict what people are going to do, or what enough people are going to do, with relative ease. Our tech-billionaire’s app knows what I might want to buy and how to sell it to me, or rather what someone like me might want to buy and there are thousands and thousands like me. Only a few hundred of us need to actually buy to make him very rich: we have; he is. Meanwhile I have an old-fashioned but stylish (I think) razor and some unnecessary but expensive socks.

But a person? Me? You? What data makes up our emotions, our feelings? What device has captured those? Is a particular emotion a specific pattern of connections between the billions of neurons in our brain? If so, where are the needles drilling into our heads to capture that data, or if not needles some other machine of loving grace?

And if it is such a pattern, what is the response to that pattern? Do I smile? Do I cry? Why? When? How will the person opposite me interpret the data I am displaying? Who and how will judge if they are ‘right or ‘wrong’, zero or one, in that interpretation? Where are their sensors, their needles?

There is simply too much data and we do not have the means of capturing it or analysing it in the moment or perhaps at all. Can we even agree what a feeling is, so we know when we’re measuring it? What’s the scale we’re going to use?

Perhaps one day AI will have sufficient art to read the mind’s construction in the face. Perhaps one day a pig will fly: after all, we understand DNA and we can genetically engineer anything: to quote the futurologist Rachel Green ‘Jurassic Park – could happen!’

It won’t though will it, however confidently it is stated. Few of us really understand AI – I don’t – and that makes us a bit scared of it as well as excited. Let’s not let that fear or excitement mislead us. In thinking about AI’s impact in the workplace, let’s go back to its basics. Without the data, without measurability of input or output, AI has nothing with which to work and therefore little to offer, not one on one.

That’s why I’m confident about the future for employment lawyers. AI can do – is doing – a first draft of any document I ask it to. Harvey is superb. Every day I find a new use for it.

Those are documents, though, not decisions. I as a lawyer have to take decisions based on hopelessly incomplete data. I try and eliminate ‘instinct’, but I cannot: I am reduced to saying to the client ‘just trust me on this’. ‘Instinct’ is built on years of experience, of watching people, of listening to them, observing how they have behaved and extrapolating from who knows what in the furthest reaches of my unconscious memory what they might do next if we pull this lever or press that button. I don’t know how the data gets in or how it gets out or even what all of the data is. We just do this stuff. Most of the time, I’m right but not always: I can’t be: no-one can. And I therefore take decisions which aim to contain the consequences of my – and other’s – predictable error.

If I conceive of employment lawyers simply as processors of data – make us the machine, if you like – then we get that data in and out through our profound empathy with both our clients and our counterparts. AI can simulate that, but never deliver it: we call it Artificial Intelligence not Artificial Empathy. That’s why here in our Gowling team we prize listening – proper, deep and empathic inquiry – as the most important of legal skills.

So, dear colleagues, we’re safe. The machine isn’t coming for us. Compare the corporate department: in M&A all deals have a price but saying all humans do is just a metaphor. Indeed, sometimes I wish if only we could buy our way out of employment claims and people didn’t care about whatever they think of as ‘justice’. What we do isn’t more difficult or valuable than what our warrior friends despatch, just more complicated; weird, even.

We deal with irrational, the ‘unfair’. We salve ‘injury to feelings’ whatever feelings are or injury to them. And ultimately every manager deals with a person, not merely a workforce. Understanding humanity means we need to understand what it is to be a human. Until the machines get a lot, lot better at capturing the data, only another human can do that.