- Everyday AI
- New York Times vs OpenAI: what everyone's overlooking 🤫
New York Times vs OpenAI: what everyone's overlooking 🤫
👀 Implications of AI and copyright, Google’s AI faces $7B lawsuit, Duolingo turns to AI, and more!
🦾How You Can Leverage:
In case you’ve been sleeping under an AI-powered rock, the New York Times vs. OpenAI/Microsoft saga has been rocking the internets.
Yeah, they wanna sue OpenAI for billions in damages for copyright infringement and legit want GPT destroyed.
(That’s what they actually requested. Lolz)
But like we talked about on today’s show, there’s some huge implications no one is talking about between the headlines.
Here’s the tl;dr version of what’s going on:
1. The NYT sued OpenAI on Dec. 27 for copyright infringement, saying they used millions of NYT articles without permission, seeking billions in damages and destruction of GPT models.
2. OpenAI finally responded yesterday, saying the NYT’s suit was without merit, and that the Times intentionally manipulated ChatGPT to get the results they submitted in their 69-page filing. Also, OpenAI argues training on copyrighted materials is fair use.
We say this every once in a while, but today’s show isn’t one you should just recap in this newsletter.
It’s worth a watch and/or listen.
The livestream audience wanted #HotTakeTuesday to be extra spicy, and I think we delivered.
Aight, y’all. Let’s break down what this lawsuit actually means, what may actually happen, and the ‘WTF how did they miss this’ moments we covered.
(Again, best you go watch/listen in full.)
We’re going FAQ style on this one.
Ya know, to spice things up. 🔥
What will actually happen?️
No one knows. The general consensus is this will get settled and won’t actually go to trial.
Why is the NYT suing?
They’re alleging OpenAI and Microsoft used millions of pieces of its copyrighted works inside of ChatGPT without attribution.
How did OpenAI respond to the lawsuit?
OpenAI clapped back pretty hard, saying the lawsuit was without merit and accusing The New York Times of intentionally manipulating ChatGPT to get the results they submitted in their filing.
They also argue that training on copyrighted material is fair use. Sheesh.
How long will this take to play out?
If it’s settled, probably not very long. But, if the case makes its way to a trial, it could take months or years to play out.
So what happens in the meantime?
If the case DOES go to court, our expectation is GenAI development will slow down as big GenAI unicorns and LLM companies shore up their legal loopholes and improve citations.
What are the big implications no one’s talking about?
Implication 1 — The New York Times and their legal team seemingly dropped the ball in its 69-page filing, which you can read here.
How did the NYT drop the ball?
By submitting screenshots of prompts and side-by-side comparisons of GPT outputs vs. the NYT copyrighted works where it was allegedly copied from.
They should have cited the sources for both parts and included shared links to the ChatGPT chats in question.
Submitting only screenshots for an LLM model output would be like a 3rd grader telling their teacher, ‘My homework is done, I swear. But I’m not going to give it to you.’
So you can get an LLM to respond any way you want?
So why didn’t the NYT include chat links?
We LITERALLY have no clue.
We thought the ‘Exhibit J’ would be the smoking gun. It wasn’t. We found the doc, as it wasn’t readily available ANYWHERE. You can go read it here.
So, does ChatGPT repeat verbatim copyrighted works?
Yes. At times it does.
It’s all about CITING/SOURCING your work y’all!
What’s the other big implication?
In the unlikely chance NYT wins this case AND their request to destroy GPT is granted, it could be potentially devastating to the U.S. economy.
Wait, they actually want GPT destroyed?
Uhhhh yeah. You can read the whole filing, or just check this screenshot.
Why would the economy be impacted?
The Magnificent 7 legit powered the U.S. economy in 2023. The Mag7 are all of the AI-powered megacap companies: Apple, Alphabet (Google), Microsoft, Amazon, Meta (FB/IG), Tesla and NVIDIA.
The S&P 500 rallied over 24% in 2023, mainly driven by the Mag7. Morningstar analysts showed the Magnificent 7 accounted for 70% of the S&P’s gains in 2023.