NYT vs. ChatGPT and the Common Crawl argument

I know a lot of people are going to want to side with the AI companies to not stifle innovation, that's kind of how we handled the early Internet. However, I think we actually need to side with the content creators to not stifle innovation. AI is only as good as its data set and AI is in the process of breaking the current internet economy where websites create worthwhile content because it gets people to visit their site where they monetize that traffic.

AI will bypass a lot of these website visits, so if we want AI to stay up to date we need a new way for people to get paid to create content, at least until the OpenAI robot influencers start making their own unboxing videos to create that data in house.

I don't think this is a partisan issue. This has the potential to be a landmark copyright case. Would you have the same reaction if it were Breitbart instead of NYT?

If the information is available to the public at various websites, then I don't see how there is copyright infringement if the source is cited the ChatGPT search results.
The AI engine is just citing the information instead of providing a bunch of links to the various sources which is a definite improvement over a conventional internet search engine result.

quote:

---

I know a lot of people are going to want to side with the AI companies to not stifle innovation, that's kind of how we handled the early Internet. However, I think we actually need to side with the content creators to not stifle innovation. AI is only as good as its data set and AI is in the process of breaking the current internet economy where websites create worthwhile content because it gets people to visit their site where they monetize that traffic.

AI will bypass a lot of these website visits, so if we want AI to stay up to date we need a new way for people to get paid to create content, at least until the OpenAI robot influencers start making their own unboxing videos to create that data in house.

---

AI art was one of the first, and because most people don't know the artists it was a "who cares" mentality while others were trying to warn that even if you didn't care about art it was a slippery slope in the worst way.

My wife is an artist, but not a digital one, but follows a few. At least one, Loish, who is pretty successful, had her art funneled into the dataset for these AI creations (without consent or compensation). It is actually important for those two points as the art is sometimes used for commercial purposes. Some people claim that it's no different than artist studying another artist, but it's apples and oranges due to the timescale involved (a person studying her art along with other artists might be good enough to compete in that space years later, and won't generate competing art with the click of a button).

It's still the same slippery slope. Just because what you do can't be automated YET doesn't mean it never can be. Soon they'll have AI that is automated to create other AI.

quote:

---

They just need to find a middle ground so the data AI needs to grow will continue to be made. Apple is currently working on an agreement with news publishers that I imagine would set some precedents in this area if they are able to strike a deal.

---

That would be good - I'm not opposed to AI getting access to content so that it can grow, but I am opposed to them freely plucking without any consideration of the creator.

Just like I'm not okay with the idea of AI taking over truck driving, but I do like the idea of AI-assisted driving.

quote:

---

This is based on the premise that we haven’t or will not reach AGI. I think we are there already and that’s what the big spectacle with Sam being ousted a few weeks back.

---

You can be the smartest computer in the world but you don't know anything if you don't have data. For example, I've seen people post ChatGPT responses where because of this lawsuit ChatGPT is not answering certain questions that might be sourced by the NYT to avoid potential infringement.

Now instead of court contested data, pretend that data never existed in the first place. How is ChatGPT 7 going to answer questions about the iPhone 18? The manufacturer produced content like user manuals and commercials only give you so much to work with. It's all of the bloggers, influencers and youtubers that are putting out content to fill in the gaps.

My comment about OpenAI robot influencers was half joking but half serious. SOMEBODY is going to have to actually experience things to tell everyone else about them. An LLM alone can't create that kind of experiential data. Pop it in a robot body and it can.

quote:

---

Pop it in a robot body and it can.

---

Do you want Skynet? Because that's how you get Skynet.