Comment by schoen

5 hours ago

With the em dashes and the conclusion "This was not called execution. It was called retirement." I expect a lot of people would believe today that these title cards were written by an LLM.

You're not wrong, but part of the core problem with LLMs is that they write every blog post and README like they're writing the opening titles for Blade Runner.

  • Also the problem with LinkedIn, local news etc. Contemporary American culture is addicted to dramatic writing and tries to leverage its tropes to create emotional engagement on everything.

What’s sort of funny about that is there’s a normal thing whenever anyone who hasn’t seen it watches Blade Runner for the first time - often they come out feeling it’s kind of derivative, when the truth is that it feels that way because everything after Blade Runner copied Blade Runner.

Add the initial title cards to that list too now, I guess.

  • It's been reported that William Gibson saw Blade Runner before releasing Neuromancer and was worried that people would think he stole ideas from Blade Runner, as the aesthetic of Blade Runner was so close to the way he imagined the Neuromancer world. Now in turn people are worried that the Apple TV adaptation of Neuromancer will feel derivative because so many subsequent works are its hereditary kin.

    What can ya do?

  • Casablanca is a famous example of this too.

    And LOTR, though in that case coming to it after reading Terry Pratchett was difficult for me (Pratchett didn't just copy its, or other, fantasy tropes, but send them up in an exceedingly clever way).

    • Yeah, there’s a bunch of these. I really like finding these sorts of “ur-“ moments, the first time a way of doing things really crystallizes into something other practitioners can see. The Matrix is another one - that movie is the transition point from 90s cinema to the 2000s, and it’s a little clunky from a 2000s viewpoint, but if you look at everything that came before it, it’s so clearly a turning point.

Thematically appropriate for it to be difficult to tell something human-created from something generated by a computer.

Almost a bit too on the nose.

Will just been seen as one of those things that just happen be to accurate to the future by mere chance.

The other was in Terminator 2 with how agreeable LLM's will be regarding new information.

Terminator/T-800 tests his foster mother's identity with a phone call.

Terminator impersonating John - "Hey Janelle, what's wrong with Wolfie?" (using a fake name instead of the dog's real name, Max).

T-1000 "Wolfie's fine," the Terminator hangs up and tells John "Your foster parents are dead."

  • If T-1000 was an LLM:

    "What's wrong with Wolfie?"

    "Wolfie's fine."

    "Actually his name's Max."

    "You're absolutely right. It's not Wolfie -- it's Max."

    "Actually we don't have a dog."

    "I apologize. I shouldn't have assumed you have a dog. I will be more careful in the future. Do you want me to list common pets in American households?"

    Terminator hangs up and tells John "You need a better model."

There's a lot of em dashes in the article as well, along with some phrasing that is very LLM. I think this was written with a mix of LLM and human.

  • What is up with the LLM police in every comment section lately… even IF so.. isn’t that the authors whole point? It can be LLM assisted, but the beauty is in the human aspects of it. This piece is, at least in part written by a human… this is the future… so?