What Quality of Language Will LLMs Converge On?

Like many professors, I've been looking uneasily at the development of Large Language Models (LLMs) and what they mean for the profession. A few weeks ago, I wrote about my concerns regarding how LLMs will affect training the next generation of writers, particularly in the inevitably-necessary stage where they're going to be kind of crummy writers.

Today I want to focus on a different question: what quality of writing are LLMs converging upon? It seems to me there are two possibilities:

1. As LLMs improve, they will continually become better and better writers, until eventually they surpass the abilities of all human writers.
2. As LLMs improve, they will more closely mimic the aggregation of all writers, and thus will not necessarily perform better than strong human writers.

If you take the Kevin Drum view that AI by definition will be able to do anything a human can do, but better, then you probably think the end game is door number one. Use chess engines as your template. As the engines improved, they got better and better at playing chess, until eventually they surpassed the capacities of even the best human players. The same thing will eventually happen with writing.

But there's another possibility. Unlike chess, writing does not have an objective end-goal to it that a machine can orient itself to. So LLMs, as I understand them, are (and I concede this is an oversimplification) souped-up text prediction programs. They take in a mountain of data in the form of pre-existing text and use it to answer the question "what is the most likely way that text would be generated in response to this prompt?"

"Most likely" is a different approach than "best". A chess engine that decided its moves based on what the aggregate community of chess players was most likely to play would be pretty good at chess -- considerably better than average, in fact, because of the wisdom of crowds. But it probably would not be better than the best chess players. (We actually got to see a version of this in the "Kasparov vs. the World" match, which was pretty cool especially given how it only could have happened in that narrow window when the internet was active but chess engines were still below human capacities. But even there -- where "the world" was actually a subset of highly engaged chess players and the inputs were guided by human experts -- Kasparov squeaked out a victory). 

I saw somewhere that LLMs are facing a crisis at the moment because the training data they're going to draw from increasingly will be ... LLM-generated content, creating not quite a death spiral but certainly the strong likelihood of stagnation. But even if the training data was all human-created, you're still getting a lot of bitter with the sweet, and the result is that the models should by design not surpass high-level human writers. When I've looked at ChatGPT 4 answers to various essay prompts, I've been increasingly impressed with them in the sense that they're topical, grammatically coherent, clearly written, and so on. But they never have flair or creativity -- they are invariably generic.

Now, this doesn't mean that LLMs won't be hugely disruptive. They will be. As I wrote before, the best analogy for LLMs may be to mass production -- it's not that they produce the highest-quality writing, it's that they dramatically lower the cost of adequate writing. The vast majority of writing does not need to be especially inspired or creative, and LLMs can do that work basically for free. But at least in their current paradigm, and assuming I understand LLMs correctly, in the immediate term they're not going to replace top-level creative writing, because even if they "improve" their improvement will only go in the direction of converging on the median.

How To Train Your Writer

Right now, on a purely technical/stylistic level, ChatGPT is an okay writer.

It's not great. But it's not bad, either. It's better (and again, we're talking purely technical here -- leaving aside factual hallucinations and the like) than some of my students, and I teach at a law school. Of course, even when I taught undergraduates I was inordinately concerned that many of my students seemingly never learned and never were taught how to write. So there has always been a cadre of students who are very smart and diligent, but just didn't really have writing in their toolkit.  And I'd say ChatGPT has now exceeded their level.

The thing that worries me most about ChatGPT, though, isn't that it's better than some of my law students. It's that it will always be better than essentially every middle schooler.

Learning to write is a process. Repetition is an important part of that process (this blog was a great asset to my writing just because it meant I was writing essentially every day for years). But part of that process is writing repeatedly even when one is not good at writing. Writing a bunch of objectively mediocre essays in middle school is how you learn to write better ones in high school and even better ones in college.

ChatGPT is going to short-circuit that scaffolding. It is one thing to say that an excellent writer in, say, high school, can still outperform ChatGPT. But how will that kid become excellent if, in the years leading up to that, they're always going to underperform a bot that could do all their homework in 35 seconds? The pressure to kick that work over to the bot will be irresistible, and we're already learning that it's difficult-to-impossible to catch. How can we get middle schoolers to spend time being bad writers when they can instantly access tools that are better?

There might be workarounds. I've heard suggestions of reverting to long-hand essay writing and more in-class assignments. There might be ways to leverage ChatGPT as a comparator -- have them write their own essay, then compare it to a AI-generated one and play spot-the-difference. I think frankly that we might also be wise to abolish grading, at least in lower-level writing oriented classes, to take away that temptation to use the bot. I don't care how conscientious you are, there aren't a lot of 14 year olds who can stand putting in hours trying to actually do their homework and then getting blown out of the water by little Cameron who popped the prompt into an LLM and 45 seconds later is back to playing Overwatch. And again, that's going to be the reality, because ChatGPT's output just is better than anything one can reasonably expect a young writer to produce.

In many ways, large language models are like any mechanism of mass production. They displace older artisans, not because their product is better -- it isn't, it's objectively worse -- but on sheer volume and accessibility. The art is worse, but it's available to the masses on the cheap.

And like with mass production, this isn't necessarily a bad thing even though it's disruptive. It's fine that many people now can, in effect, be "okay writers" essentially for free. It's like mass-produced clothing -- yes, most people's t-shirts are of lower-quality than a bespoke Italian suit, but that's okay because now most people can afford a bunch of t-shirts that are of acceptable quality (albeit far less good than a bespoke Italian suit). The alternative was never "everyone gets an entire wardrobe of bespoke Italian suits", it was "a couple of people enjoy the benefits of intense luxury and most people get scraps." Likewise, I'm not so naive as to think that most people in absence of ChatGPT would have become great writers. So this is a net benefit -- it brings acceptable-level writing to the masses.

If that was all that happened -- the big middle gets expanded access to cheap, okay writing, with "artisanal" great writing remaining costly and being reserved for the "elite" -- it might not be that bad. But the question is whether this process will inevitably short-circuit the development of great writers. You have to pass through a long period of being a crummy writer before you become a good or great writer. Who is still going to do that when adequacy is so easily at hand?

I'm not tempted to use ChatGPT because even though my writing takes longer, I'm confident that at the end my work product will be better. But that's only true because I spent a long time writing terribly. Luckily for me, I didn't have an alternative. Kids these days? They absolutely have an alternative. It's going to be very hard to get them to pass that up.

I'm That Guy! I'm That "Independent" Artist Guy!

You know the character. In a TV series or a movie, he's a talented artistic sort, but utterly rejects any commercialization of his product if it entails compromise in his work. Using his song for an advertising jingle? No way. Altering his masterpiece to suit the tastes of the wealthy client/benefactor? Absolutely not. He's fiercely independent, and disdains the trappings of popularity or financial reward if it in any way interferes with the sanctity of his art.

I always hated those characters. I found them sanctimonious, self-absorbed, and more than a little ridiculous.

Except I just realized ... I might be that guy, with respect to my writing.

Of course, I'm not a TV or movie character, so I'm a more filed-off version of the sort. I'm willing to publish my opinions in the popular press, and I don't harbor any particularly negative opinions about gaining popularity or exposure or money (I'll let the readers decide whether I'm "talented" or not).

But it is certainly not why I write, and I'm pretty consistent in rejecting opportunities for greater exposure if I have even the slightest concern that the medium will entail a dilution or compromise in my message. Which it usually will, since writing for someone else involves incorporating their editorial judgment and word count restrictions and other formatting strictures. Whereas when I write on this blog, where I can write for as long or as short as I want, on whatever topic I want, for whoever cares to read (or doesn't) -- none of that applies. I'm in control of my own message.

The same applies to other potential communicative mediums I might join in. Interviewed on a TV series? Someone else is the editor of the clip -- that makes me very nervous. Serving as an expert witness? You're supposed to give decisive judgments about whether X or Y did or didn't happen -- but what if I want to equivocate? This is also why I've never had an interest in going into politics: the arguments that win elections or change popular opinion aren't necessarily good arguments, and I don't have the stomach for making claims that rouse popular support even as I know they're not analytically precise.

Of course, these self-imposed limits mean that fewer people will be exposed to the arguments I make, and one might think that -- given all the energy I invest in crafting them -- I'd view that as a sizable loss. But ... I don't. Not really, anyway. I'm always happy when people read what I write and find it compelling or persuasive, but that's never been my primary motivation. I'm a very inwardly-motivated individual; I write because I want to get my thoughts out into the world. If they happen to spark some other person's imagination (or get me some freelance dollars), that's great, but it's just a bonus.

I'd like to think I'm not sanctimonious about the whole deal: I don't think I'm better or purer or more authentic because I've made the choices I've made. But I'm definitely making that style of choice. And the more I think about it, the more the closest archetype I'm falling into is one of those independent artist types. I am, it turns out, That Guy.

Almost Floridian Roundup

On Saturday, Jill and I depart for our annual holiday trip to Florida (her family winters in the panhandle, and mine has begun the traditional Jewish relocation to Boca Raton). We'll be there for about a week before heading up to Boston for New Year's.

* * *

Given the ongoing tensions between "Bernie Bros" and racial minorities, this article on the DSA's race problems is wholly unsurprising (I know some East Bay DSA folk, and they -- the organization, not the folks I know -- have a gender problem too).

Netanyahu is frantically trying to stop Trump from releasing his Mideast Peace plan (which sounds like it is basically ... the Clinton peace plan), worrying it will make him vulnerable to a right-wing challenge. This is torture for me, because normally a Bibi-Trump fight is a "root for injuries" situation, but here I actually have to be constructive and want whatever makes a just peace agreement most viable (and I honestly don't know if "releasing it now" or "later" is more likely).

I had been meaning to link to this outstanding meditation by Erika Dreifus on being a Jewish-American writer in 2018, and just never found the right moment. Well, no moment like the present.

This is an okay -- not great, not terrible -- article on how toxic masculinity intersects with Zionism. I think I did a better version, though.

Albania expels Iranian diplomat after alleged plot to attack an Israeli soccer match.

When I read that the Women's March was consulting with "Jewish groups" on the question of antisemitism, I just assumed it would be JVP-style groups. But I was wrong, and I'll give due credit: the NCJW, Bend the Arc, and JFREJ represent a fair spectrum of progressive Jewish organizations to give perspective on these matters.

How To Write a Brief, By Paul Clement

This amicus brief by Paul Clement on behalf of the American Beverage Association (the other, other ABA) in POM Wonderful v. Coca-Cola is a masterpiece. The formal legal question -- on which I express no opinion -- is whether a party can bring a suit under the Lanham Act (which governs false advertising) regarding a label of a product regulated under the Food, Drug, and Cosmetic Act. POM's lawsuit accuses Coca-Cola of mislabeling a drink as "Pomegranate Blueberry" juice when it contains just .3% pomegranate juice and .2% blueberry juice (it is 99% apple and grape juice).

And when you put it that way, it sounds really bad! There's an old saying amongst lawyers that when the law is on your side, pound on the law, and when the facts are on your side, pound on the facts (and when neither is on your side, pound on the table). These facts sound very bad for Coca-Cola (99% juices other than the two prominently displayed on the label), so the conventional wisdom would be to bang on the law and downplay the facts.

My theory of advocacy, however, has always been the opposite -- go on offense where your opponent thinks they're strongest. "Offense" is important -- the strategy only works if you argue that the point they think is their ace really, truly, obviously favors your own side. Even if you don't persuade the audience, it has a neutralizing effect -- it makes it seem like the issue is at best a single or a double for the other side rather than they're home run.

I digress. The point I want to make is that Clement starts with what seem to be awful facts, and deftly explains why labeling juices from the perspective of volume may not make a lot of sense. It is clear, lucid, and intuitive -- no small feat given how incongruous the position seems at first glance.

If one is looking for a model of how to write a brief, this is one to do it. It is a masterclass of the art.

This is How I Vacation

I'm taking my first vacation from work next week (week of February 10). I'm taking vacation now in part because its my birthday on February 11 (and who wants to celebrate in the office), and in part because I didn't take vacation over the holidays because I had volunteered to work on a bunch of cases with filings on or close to New Year's Day. In the immediate term, this means that I have a ton of work to do this week, since I'm trying to get it in under the wire. But nonetheless, it is very exciting -- I'm looking forward to having the week off.

People have asked me "where [I am] going" for my vacation, to which I have answered "my bed." I don't like traveling much to begin with, and in the context of a vacation I view it as basically sacrificing at least a half day if not more for no discernable benefit. I can sleep in and read books in my apartment just fine. Which leads to the second primary way I've prepared for my vacation: buying five books on deliberative democracy, which I will use to help write a new law review article I'm working on.

I haven't mentioned that part of the vacation plan at the office yet, but if I did I'm sure they'd look at me like I'm crazy. "You're taking a vacation from being a lawyer by doing a bunch of legal (or law-related) reading and writing?" This response makes superficial sense, but I honestly can't parse it. I understand why someone might not be interested in doing academic research and writing for fun, but they of all people should know that such work is a far cry from our day-to-day legal work. This is not a postman going for a walk on his day off.

I'm reminded of a story from when I was a summer associate at the firm. I was at some after-work event (I think it was basketball), and I had brought a book to read during the downtime -- a retrospective honoring the career of Iris Marion Young. A well-meaning associate saw me reading the book and said, concerned, "I hope you're not working tonight!" Now, of course, he was absolutely right that a summer associate should not have to be working through a Friday night after-work firm event -- that's the whole point of being a summer! And I told him no, this book was just for pleasure. But I wanted to say "buddy, if there's a practice group here where I might plausibly be reading books on contemporary feminist theory and democratic practice for work, by all means direct me to them right now because I need to get on that train."

In any event, writing is how I relax. Reading is how I relax. Firm life gives a lot of opportunities to read and write, of course, but not always in the forms and topics one might like. So for my vacation, I'm going to read what I want to read and write what I want to write, and I assure you I will enjoy the break thoroughly.

It Bears Repeating

Yesterday, I commented on how Ta-Nehisi Coates' site is one of the very few with a large, worthwhile comments section. Today, I will comment on a very different aspect of his site -- that he is a damn good writer. I'm envious.

Sotomayor Too Thorough

Apparently that's literally the latest complaint about her: That her opinions are too methodical, address too many issues, and don't have enough quotable catch-phrases. Of course, everyone knows that thoroughly addressing each and every argument is a hallmark of judges who lack the intelligence to be on the bench. [/sarcasm]

As an aspiring attorney, it is certainly more pleasurable for me to read opinions that have a little snap to them. But a judge's first job isn't to be a stand-up comic, it is to, well, calmly and methodically dispose of the issues before them. And I will note that judges who wrote for the quote sometimes miss the important legal issues in the process. So once again, advantage Sotomayor.

Pace Yourself

I tend to write in spurts. I'll spend long periods of time doing absolutely nothing, but if I hit a rhythm I can have tremendous bursts of productivity. However, right now I'm working on an article that I really like, and I'm having a lot of trouble getting words on the page. The problem is that whenever I have one of those flashes of inspiration, I get really excited and can't focus on actually writing it. Instead, I just pace around the house talking myself through it, and I don't actually write anything down. By the time I've calmed down enough to sit back in front of the computer, the moment is gone, and I still can't write.

Does this (or something similar) happen to anyone else?