Template:M intro work Large Learning Model

LLM
(n.)
LLM Once meant a “Master of Laws” postgraduate degree, and was the mark of a sensei in the ninja brotherhood of legal services. Now all one needs, we are told, for that kind of expertise is a different kind of LLM — a large language model. A form of artificially intelligent chatbot. The legal profession is to ChatGPT, we hear, as poor Chrissie Watkins was to Jaws.

But the JC is not convinced, for those making this prediction do not ask the question: “cui bono?” In the annals of legal practice the person who benefits first, and often last, is the lawyer.

Now.

It is a truism that she who has a tool uses it, firstly, for her own benefit.

Acommercial lawyer’s business model is predicated on two things:

(1) time.

(2) ineffability.

It is a happy accident that, generally, (2) begets (1). As is the fact that the more ineffable something is, the longer it takes. The longer it takes, the more you a lawyer can charge. Commercial legal contracts take a long time. (Hence, no commercial law firm on the planet really cares for plain English. Oh, they all say, they do, of course, but come on. Have you ever read law firm boilerplate?)

But here’s the thing: it will be lawyers who will use ChatGPT as a tool, not their clients. Why? Because ChatGPT is a pattern-matching device. It understands nothing. It cannot provide unmediated legal advice. It is a “back-breaker”: the “last mile” needs a human who knows what she is doing. Nor, importantly, can ChatGPT write legal opinions — well, meaningful ones — and nor does it have professional indemnity insurance or the deep pockets it would need should a client ever want to sue it.

It is not plausible that ChatGPT will be deployed, ever, without the intermediation of someone who knows what she is doing. That someone will be a lawyer.

Now a “last mile” lawyer could use ChatGPT to simplify documents, accelerate research and break legal problems down to significant essences, thereby reducing the cost, and increasing the value, of her service to her clients. She could give all this value creation up for nothing. Or she could use ChatGPT to further complicate documents, convolute language, invent new options, create new contingencies: to build infinitesimal detail into her work that it was just too hard to do manually.

Which, realistically, should we expect a self-respecting lawyer to do? Simplify, or complicate? To sacrifice time and ineffability, for the betterment of her clients and the general better comprehension of the unspecialised world? Or would she plough her energy into using this magical new tool generate more convolution, ineffability, and recorded time?

Rememberr: for 40 years we’ve had technology — Microsoft Word, mainly — which could have been used, powerfully, to simplify and minimise legal work product. Did any of them do that.

You can already see the effect LLMs are having on legal work product. NDAs are getting longer, and worse.

Chat GPT may disrupt a lot of things, but it won’t be disrupting the legal profession any time soon.

Bear in mind who ChatGPT would be disrupting in this case. Two things about consumers of high-end commercial legal services.

(1) Most of them — us — are lawyers

(2) As lawyers they — we — take pride in the ability to work with difficult, complicated things. Convolution is a measure of our worth. The love of convolution for its own sake, for what it says about us, is a strong common value between lawyers and their commercial clients.

Lawyers — inhouse or out — are the jazz aficionados of text; cineastes of syntax. Overwrought contracts are expected: nothing says “prudent management of existential risk” like forty page of 10pt Times Roman. Plain language is not for serious people.

That is to say, neither fee-earning lawyers nor their immediate clients want plain contracts. If they did, we would already have them.