Template:M intro work Large Learning Model: Difference between revisions

''[[LLM]]s work the same way''.  

Like all good conjuring tricks, [[generative AI]] relies on misdirection: its singular genius is that it lets us misdirect ''ourselves''. We wilfully suspend disbelief, never noticing who is creatively re-rendering machine-made screed as magic. ''We are''. We are neuro-linguistically programming ''ourselves'' to be wowed by a {{plainlink|https://en.wikipedia.org/wiki/Digesting_Duck|mechanical duck}}, when the clever part is really happening inside our own heads. Humans are seldom stupider than when they misattributing their own genius to a box of gears and pulleys.

By writing “prompts”, we create an expectation of what we will see. When the pattern-matching machine produces something roughly like that, we use our own imaginations to backfill, frame, filter, correct, boost, render, sharpen and polish the output into what we ''wanted'' to see. We ''construe'' the output to conform to our original instructions.  

When we say, “fetch me a tennis racquet”, and the machine comes back with something more like a lacrosse stick, we are far more impressed than we would be of a human who did the same thing: we would think such a human a bit ''dim''. But with [[generative AI]] we don’t, at first, even notice we are not getting what we asked for. We might think, “oh, nice try!” or “well, that will do,” or perhaps, “ok, computer: try again, but make the basket bigger, the handle shorter, and tighten up the net.” We can iterate this way until we have what we want — though note all the refining intelligence is coming from ''us'' — or we could just Google for a conventional photo of a tennis racquet. As the novelty wears off, that is what more and more people will do.

Either way, as we persevere, we see more and more of the magician’s wires. We sense how the model works: its familiar tropes, tics and consistent ways of doing things, which are never quite what you had in mind. The first couple of passes are great: they get from zero to 50% of what we wanted. But the marginal improvement in each following round diminishes. The piquant surprise at what the machine can produce dampens at each go-round, reaching asymptotically towards an upper capability, well short of 100%, in “producing what you had in mind”. Eventually we settling into an [[Entropy|entropic]] and vaguely dissatisfying quotidian. “Okay, it isn’t quite what I had in mind. but in the interests of time it will do.”   

Again, we subordinate our own interests to the machine’s. We accept a mediocre job we would not take from a human. We must be more exacting lest we let the machines become our overlords, not by conquest but our own pathetic surrender.

Ours is the skill going in, and ours is the skill construing the output. The machine does the boring bit:  the bit that machines have ''always'' done: applies uninspired strength, speed, reliability, and economy. ''Not ingenuity''.

Now, in all kinds of literature ''bar one'', “reading” is where the real magic happens. ''Construal''. It is the [[Emergent|emergent]] creative act and community consensus that renders ''Hamlet'' a timeless cultural leviathan and {{br|Dracula: The Undead}} forgettable pap.<ref>Maybe not ''that'' forgettable, come to think of it: it has stayed with me 15 years, after all.</ref> A literary work may start with the text, but it stays there barely a moment. The “meaning” of a literary work is ''necessarily personal'' to the reader: it lives between the reader’s ears, and within the cultural milieu that interconnects the reading population over the generations — ''all as construed by the apparatus between the reader’s ears''.

There is one kind of “literature” that ''is'' like a computer programme: where the ''last'' thing the writer wants is for the reader use her imagination, , or bring her cultural baggage in from the hall and use it to “construct” a meaning. In this singular domain, clarity of the writer’s intention is paramount: the only priority is divining the what those who wrote the text meant by it.

Now: the punchline. Given how integral the reader and her cultural baggage are to the creative act in ''normal'' literature, we can see how, in that domain, a [[large learning model]], which spits out text ripe with interpretative possibilities, which positively begs for someone to “construct” it, just to make sense of it, is a feasible model for the language of possibility. The interpretative task is paramount: to move from a model where ''most'' of the creative work is done by the reader to one where ''all'' of it is, is no great step.

 

''But that does not work for legal language''. Legal language is code: it must say exactly what the parties require: nothing more or less, and it must do it in a way that leaves nothing open to later creative interpretation. Legal drafting is as close to computer code as natural language gets: a form of symbol processing where the meaning resides wholly within and is fully limited by the text.

But unlike computer code, the operating system it is written for is not a closed logical system, and even the best-laid code can still run amok. You can’t run it in a sandbox to see if it works. You have to test in production. A random-word generator that creates plausible sounding legal text, which falls apart on closer inspection, is not much use.

This is not to say that a [[large language model]] can’t be used to generate legal [[boilerplate]]: it just can’t do it by itself, and the process of working with it will be a lot more labour-intensive than the first round of generation suggests.  

We have seen it suggested that one might invert the process instead, letting humans do the first cut, then pressing the machine into action to critique the human drafts, to find potential errors and omissions.  

Boilerplate is [[boilerplate]] for a reason. It is pinned down; tried, tested and done: it takes those things that should go without saying out of the equation. There is nothing to be gained from having a [[large language model]] drafting boilerplate from scratch each time — especially if, randomly, it changes things. ''[[The quotidian is a utility, not an asset|Boilerplate is a utility, not an asset]]''. It is the part of the deal the lawyers already spend the least time on. The parts that matter — that they do spend time on — are typically discrete, manageable parts of drafting. It is hard to see how an LLM really assists in the process.