Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

FWIW, we at Medium feel pretty similarly to Reddit but with a yes to this question about whether authors should get paid.

AI companies are betraying basic business principles: they are taking value from datasets like Reddit and Medium without giving any value back. Fine if you can get away with it. But since AI, especially text based LLMs, relies on source material, it's pretty straightforward for the platforms that host that source material to deny access. Things like ChatGPT do need current source material.

I don't think it'll come to a war though and that the AI companies will instead give some value back. It could be as simple as citations that send traffic back. That's essentially the exchange of value that we all have with Google these days.

But if it's money, then I think the obligation is for platforms to pass that on the authors. It'd be hard for an individual author to negotiate this on their own with a company like OpenAI, but platforms are in a good position to negotiate on their behalf.



The content is public. Internet companies are staring to sound like Disney.

The AI is definitely giving value back.


It’s public for humans, not for other businesses stealing it for their own profit. I don’t want to be an anymous contributor to an AI.


Another in a really long list of issues that need a clear differentiation between human consumption and machines. So many things that were innocuous or even useful before the age of ubiquitous cameras, other sensors, and computers, are now a big problem.


As an end user I get a lot of value from AI companies.

I’m curious how a content platforms TOS will matchup against a Search Engine’s webcrawler TOS.

“I want people to find and access my content, but I own it.”

vs

“I will send people to your content, but any public data I can access, I can store and process how I want.”


You're in no position to negotiate this with OpenAI because they already have the relevant data stored locally. So does Google/Bing. You could be in a position to negotiate it with smaller upcoming OpenAI competitors, but all that will achieve is granting OpenAI and Google/Bing a monopoly because their competitors will have new large costs that they don't.

Also, Medium has a metered paywall already. Why not just let them open up a corporate account and pay to access paywalled content the same way users do? Why are any negotiations required?

BTW I use Medium but I never use the paywall. I'm fine with my content being used to train AI for free. The payments and tax complexity involved aren't worth the tiny amount of income that any such deal might generate, nor do I want OpenAI to have a monopoly.


Maybe no position with Google because they can bundle it with search results and threaten to take away search traffic. But OpenAI definitely does not already have all the relevant data. They need the new stuff also. That's part of the Reddit position as well.


I wonder to what extent that is true, now LLMs can search the web and read the results?




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: