Multifaceted: The linguistic echo chambers of LLMs

This is a fun one.

I’ve spent more time than I’d care to admit staring at LLM output. And there’s something that I’ve noticed: LLM-generated prose has a kind of… vibe. It’s difficult to describe, but in this initial era of LLMs, it tends to be fairly obvious when you’re reading an AI-generated piece of prose.

One giveaway I've noticed is this particular turn of phrase:

“Culture is a complex and multifaceted ...”
“Intelligence is complex and multifaceted ...”
“Technology is a complex and multifaceted ...”

In the true Dawkinsian sense, the phrase 'complex and multifaceted' has become a meme. I've seen it again and again in outputs from GPT, but to double-check, I did a bunch of GPT-3.5 generations (code here). Here's what I found when generating completions for a prompt of 'complex and ...':

There is a bizarre prevalence of the term 'multifaceted' specifically. Why?

I wanted to understand whether this phrase and the specific word 'multifaceted' was newly popular or had existed for a while. As a first port of call, I had a look at Google Trends. And I observed a very shocking increase within the last year:

At this point I wanted to get an indication of whether this was an online-only trend. It's hard to establish this but I thought I'd try Google Books' N-gram viewer. Maybe it would show me. And, as suspected, we see no notable inflection, although one can see there's a gentle increase over time.

Tangent: For what it's worth, I find it a bit of a weird phrase. It's a tautology, as 'complex' and 'multifaceted' are almost synonomous. It reminds me of legal doublets like 'null and void' and 'cease and desist'. It's a rather nice and affirmatory way of saying something. I guess it sounds clever and informed, which is, after all, the vibe LLMs are going for.

Anyway, I wanted to go a bit further in order to ensure this was actually a newly prevalent phrase online. Google Trends isn't very convincing by itself. So I went digging for other places where linguistic…