Hacker News Reader

Comment by meerita | Hacker News Reader

> Two neighbors stop in front of a construction site and look through the fence. Inside, five workers are using two enormous machines to finish digging a massive hole. One neighbor shakes his head. “Fifty people could have dug that with shovels. Everyone would have had a job.” The other looks at him and says, “Why stop at fifty? A thousand people could have dug it with their bare hands.”

Good luck trying to figure out whether something was written by AI or a human. This mass hysteria around AI will soon be forgotten. Nobody will care whether something was written by an LLM or a human. People will just move on to other topics.

Downvoting me or others will not change this fact.

View on HN

I don't think I've ever encountered something written by an LLM that was better than barely passable prose. It's just about suitable for documentation and mundane news regurgitation, but a piece of creative writing? Please show me your favourite examples where it's better than "passable", I'm open to changing my mind.

Figuring out whether something is written by AI or a human is fairly easy with Pangram, and soon all major models will embed fingerprinting and it will be basically 100% detectable.

That's a consequence of not using enough compute. The correct way to have an LLM write creatively is to have it try hundreds of permutations and then judge those hundreds. This is basically what humans do anyway. Some models are better than others but you can absolutely write well with it. I generally find that when people say it doesn't work for them they aren't actually trying very hard or they simply do not know what they are doing.

A personal account doesn't change the fact that AI will become so ubiquitous that most people won't be able to tell for sure whether something was written by a model, let alone a model following specific instructions like "Write like Mark Twain."

I'm waiting for concrete examples.

The second I realize that any kind of art was made by AI, I lose all interest in it, because "made by AI" is the same thing as made by no one. Some of us will always care.

What will happen when you cannot trust your instinct to see if something was handmade or made by a machine? Will you give up your bias or will you go full bias just in case?

Yep, same. AI interesting only in the same way that a forklift is interesting. It's technically remarkable, it can do more work faster than a person, but without much finesse and certainly without any soul. The entire point of art is that it's an expression of another human. AI can never create art by definition.

The value of moving dirt is that someone paid for a hole. They don't care how it was made.

The value of written communication is in whether people decide to read it. There is nothing controversial about readers rejecting content they perceive is low-effort.

Humans can tell if something is written by an AI. LLMs, so far, have yet to pass the uncanny valley, which is probably part of why we all react with such disgust when LLM-generated content is posted to HN.

> Nobody will care whether something was written by an LLM or a human. People will just move on to other topics.

Do you have a crystal ball that tells you this, or is it just your gut feeling? Last I checked, AI and AI-generated content continues to be deeply unpopular outside of insular tech circles.

You can already see the effect of coding. More and more people care less if something was the output of a LLM or a human who spent a day writing it.

It's pretty obvious to me when someone just copy pastes claude spew onto a website and calls it a day. If they put in enough effort that I can no longer tell then sure, I will read that.

Yeah it's pretty obvious now. But look at the improvements Anthropic made to Opus 5.5. It went from physically painful to read, to kind of annoying but tolerable. I wouldn't bet on it taking a long time to reach high quality prose.

Btw it is kind of weird how it still can't write prose normally but it can write impressively high quality code now. Must be something to do with their training. Or maybe it's easier to quantify good code.

Just a few more hundred billion and it’ll be good, trust me bro.

Uhm yeah probably. Are you suggesting that LLMs have been stagnant despite investment?

I don't understand your point. They're getting better so quickly it's hard to keep up.

And this time next year I'll have 365 wives, and my CPU will run at 2 terrahertz.

Development and improvement is often very non-linear, assuming you can extrapolate forever is just as much guesswork as claiming it'll never change.

Personal perception will never be a reliable way to detect text written by LLMs, especially when people guide those LLMs with specific rules and instructions. It's virtually impossible. What you'll experience instead is your bias growing to the point where nothing will seem human-written anymore. You'll start assuming all text was generated by AI. It's pure AI paranoia.

> Good luck trying to figure out whether something was written by AI or a human.

It’s a magazine, not a social media website. They control who writes for them and they know who they can trust. If any of those people break that trust, they are banned and replaced with someone else.

> Nobody will care whether something was written by an LLM or a human.

Of course they will. To this day people still care about (for example) colonialist actions from hundreds of years ago or buying print books. It’s disconnected from reality to believe everybody will simply not care who wrote what they read.

"It's a magazine" is not a bulletproof solution to avoid AI contributions. I worked for many magazines in my younger years, back when people worked and wrote in the same place. Now everything is digital, and writers are spread all around the globe. They share information digitally, not through typewriters. These people who are suddenly concerned about AI have no reliable way to prevent or detect AI-generated text anymore.

> "It's a magazine" is not a bulletproof solution to avoid AI contributions.

Nor have I claimed such a thing. I recommend you read the rest of the paragraph instead of stopping at the first sentence and making a bad faith argument. If you don’t read a point to the end, you can’t understand it enough to argue against it.

My friend, you used this as an argument: "It’s a magazine, not a social media website. They control who writes for them and they know who they can trust."

That is not a bulletproof system. Human trust is not deterministic, it's not precise, and it's based on biases. That's the point.

Damn, and you’re still not reading to the end. I’ll help. Here’s the rest of the paragraph:

> If any of those people break that trust, they are banned and replaced with someone else.

Do you understand what that means? I’ll help again. I means that I never claimed it was a “bulletproof system”, as you keep insisting. In fact, that makes it pretty explicit that I know it’s not bulletproof, because I exposed a failure case!

Here’s the thing, though: It doesn’t have to be bulletproof, and it is absurd to believe it does. It just has to be good enough, and you can get good enough by trusting people and evaluating on a case-by-case basis. Because you know what? Not everyone is an asshole who wants to fool others. The only people who believe everyone else is untrustworthy are the people who are untrustworthy themselves.

But you cannot enforce a deterministic way to know whether someone has broken that trust, in fact, if it is about feelings, that can fail or hit the truth. It will be transparent to everyone, but impossible to know if someone has broken the rules. That's the main point. Of course, most people are trustworthy, but who knows? Maybe one day someone wakes up and decides to break those rules, and no one will ever know.

>his mass hysteria around AI will soon be forgotten. Nobody will care whether something was written by an LLM or a human. People will just move on to other topics.

Just like they've forgotten about tulip bulbs?

Even if LLMs deliver a fraction of what their advocates promise, their impact entry into society will be relevant for centuries, much as with piracy, industrialization, and abolition.

Reply on news.ycombinator.com