Remix.run Logo
conception 16 hours ago

The problem is LLM output’s value is intrinsically zero because there is no scarcity to it. I too can just prompt “What’s the rate of self hosted email?”. If an article is signaling that its value is zero, then everyone should just move on as it’s not providing anything to anyone. Effort is the economies main signal of value, good boots that last cost more in the classic example. The only time it’s not zero is when there’s a shared understanding between those consuming the output - a company wanting your llm code, friends making funny images for an inside joke to share, etc. And some people grouped may be specifically interested in AI output - people sharing music on Suno. But in my experience no one cares about your AI generated song, image, article, app if it appears to just be my own prompt away from me having it too. Give a significant other an AI image printed out for an anniversary gift - see how excited they are about it. Versus getting an artist or photographer to do the same. As Claude might say, it’s not laziness, it’s effort.

tzs 10 hours ago | parent | next [-]

> The problem is LLM output’s value is intrinsically zero because there is no scarcity to it. I too can just prompt “What’s the rate of self hosted email?”. If an article is signaling that its value is zero, then everyone should just move on as it’s not providing anything to anyone.

That only applies when the article's author just prompted an LLM (and did not supply the LLM with any data you did not have) and then posted the results.

I'd expect most article authors who use LLMs are more involved then that, checking the results and correcting (or telling the LLM to correct) thing that it got wrong.

If the article is from someone who knows more about the subject then you do then that will be more valuable that what you would get asking the LLM yourself (even if the prompt used in the article didn't include anything that your LLM didn't know).

conception 7 hours ago | parent [-]

Sure but if they aren’t editing their copy to make it readable then I probably won’t trust anything else. It’s not hard to read what comes out of an LLM and make it readable.

jstummbillig 15 hours ago | parent | prev [-]

> The problem is LLM output’s value is intrinsically zero because there is no scarcity to it.

I think if this post had not gone the AI-generated-text-route, some people here would easily have considered it interesting for its content, even if it was entirely AI generated.

conception 15 hours ago | parent [-]

That’s the worst part if they had just spent 10 minutes editing it it would’ve been fine but either they couldn’t even put in 10 minutes of effort into the article (bad) or they can’t tell that it’s written obviously like an LLM (also bad). Neither outcome makes me want to read the article more.