[Update 13 May 2017: I should make it clear that I’m not suggesting that automated contract drafting is by itself enough to make one an informed drafter. Instead, my point is that comparing the two means of generating the bulk of contract text, automated contract drafting is a better vehicle for training than is copy-and-pasting. But by itself it’s not enough. You’d also need MSCD and whatever reliable subject-matter authorities are out there. And a mentor of some sort would be valuable too.]
Via Noah Weisberg (@nwaisb), I learned of this article on Quartz by Sarah Kessler. It’s entitled AI Will Rob Companies of the Best Training Tool They Have: Grunt Work. Here are the most relevant bits, for our purposes:
Here’s the problem, Engelbert says: “Where do [those middle-level employees] get that experience and judgment? That’s probably the number one thing I worry about as we shift our model.”
…
Reviewing endless contracts may may be tedious, but it’s also instructive. And without this entry-level experience, Deloitte will have to find new ways to train and develop middle-level employees. Companies across other industries will face a similar problem.
…
… All that grunt work is actually instructive for workers. And without any experience in the trenches to draw upon, that training will be hard to replicate.
That might be the case for contract review, but I suggest that it doesn’t apply to contract drafting. If we’re presented with a glorious future in which we draft contracts by answering an annotated online questionnaire (something that, by the way, has nothing to do with AI), no one should mourn the good old days of copy-and-paste.
Copy-and-pasting involves recycling, largely on faith, precedent contracts of questionable quality and relevance. You generally have little opportunity to learn much about the dysfunctional language you’re recycling, but you become unhealthily habituated to it. It’s all you know. And much of what you do learn is likely based on flimsy conventional wisdom.
So there’s little reward in all the drudgery that goes with cranking out drafts the copy-and-paste way. You’d learn way more by consulting the guidance offered in an annotated document-assembly questionnaire produced by someone reliable. You’d learn still more by consulting an output document containing high-quality contract prose revised to reflect answers selected in the questionnaire.
That sort of rigorous, concentrated training is also more systematic than most of what passes for mentoring these days.
You may be creating a false dichotomy here. There’s a difference between learning well and learning badly, but given the dearth of law school instruction on contract drafting (both the how and the why), learning by doing is the only available alternative. Tapping things out on a computer or calculator may give you an answer, but you won’t know whether to question the answer unless you’ve learned long division and even long root extraction.
An automated contract might be “perfect” within certain parameters, but relying just on that, without knowing why the program created the language it did and put it where it did, will ill-equip a lawyer to respond when a counter-party questions the output. “Because reasons” is generally not a persuasive response. Heck, you even wrote a book about *why* certain kinds of contracts are structured the way they are, so you obviously think there’s more to learn than just what the Contract Express Oracle thinks a clause ought to say.
So, until law schools overcome their litigation bias in teaching the law and the skills of legal practice and give negotiating and drafting agreements their due, the apprentice system is what we’re left with. Short-circuiting it will not produce better lawyers.
In my scenario, the automated template offers two advantages. First, it uses MSCD-compliant language. And second, it explains choices and allows you to see the changes made to the document to reflect those choices.
Normally you’d provide annotations only relating to choices; you wouldn’t explain what’s behind all the other wording. For that, you’d have to turn to MSCD or other authorities. Doing the copy-and-paste thing certainly wouldn’t provide you that information either.
In my view, becoming a good contract drafter requires lots of practice – writing clauses, correcting them, having defects pointed out by others, debating the wording, looking at how others have drafted similar clauses. Using automated drafting might provide some of that experience, but I suspect it won’t produce rounded drafters who can draft non-standard clauses. In my world there are lots of non-standard clauses, but I can see that some areas of practice may lend themselves more to automation.
Of course. But what I’m getting at here is how you acquire a grasp of the basic vocabulary. Parroting traditional language won’t do it.
Sure. But nor, in my view, will studying the guidance in an automated template. It will help, just as studying the Highway Code helps you to drive better, but only once you have mastered the basics.
I could try stretching the analogy to learning to drive in an automatic car won’t help you with manual gears, but that may not translate across the Atlantic!
I acknowledged as much in my response to Vance. Automation is just one element of informed drafting.
I also dream an automated system. Such system is fantastic, especially a system that may generate contract of different language simultaneously. It is highly probable, although some effort made by UN e-business group do not have positive effect. I think an automated system that works needs to care four dimensions. The first is basic language of a contract (declaration, rep.&warranty, covenant, …..). The second is parts of a contract (from title to signature). The third is business function to be done by a contract (flows of product, human service, information, money and titleship). The fourth is transaction timeline.(such as in long term supply of goods, from sample providing, design-in, forecast, order, delivery, return/acceptance, to payment, repair/maintenance). By considering these dimension, different contract modules can be designed. Maybe the better solution is not through sample sales agreements, distribution agreements, but through sample money flow modules, sample product flow modules, and other sample boilerplates.