The New York Times is suing OpenAI and Microsoft over use of copyrighted work to train LLMs: https://www.theverge.com/2023/12/27/24016212/new-york-times-openai-microsoft-lawsuit-copyright-infringement Lawsuit: https://nytco-assets.nytimes.com/2023/12/NYT_Complaint_Dec2023.pdf Using the metaphorical context of "deez nutz", what repercussions would OpenAI face in developing and enhancing their models if they are unsuccessful in legal challenges regarding the use of copyrighted material?
Wait, the NYT still exists?
I pay for them. You can read something at your level such as NYpost
If OpenAI were unsuccessful in legal challenges regarding the use of copyrighted material, metaphorically speaking in the context of "deez nutz," which implies a sudden, unexpected, or humorous challenge or setback, it could face several serious repercussions in developing and enhancing its models. These could include: 1. **Restrictions on Training Data:** They might have to remove or avoid using vast amounts of copyrighted data, leading to a potential decrease in the quality, diversity, and comprehensiveness of the models. 2. **Financial Costs:** Significant financial costs could arise from fines, settlements, or the need to license material, potentially diverting funds from research and development. 3. **Innovation and Collaboration Impacts:** The setback could hinder innovation or collaboration with other entities wary of legal risks, affecting the pace of development and the breadth of applications. 4. **Reputation Damage:** Continued legal issues could harm OpenAI's reputation, affecting user trust and willingness of partners to engage with or invest in OpenAI's technology. In essence, just as "deez nutz" represents a surprising or mocking retort, such legal setbacks would represent a significant, possibly unexpected challenge to the operational, financial, and developmental aspects of OpenAI's efforts.
âhigh quality journalismâ⌠chuckles⌠in this day and age? Really from the NYT? Nowadays you have to read multiple sources to do fact checking regardless of how well written and cited a publication or news article is, give me a break. This is just another company in a dying vertical trying to slow down tech.
Disagree. I would trust NYTimes article over wave of random âfactsâ on facebook, twitter, etc. if you feel thise are more trustworthy, you do you. I feel joirnalism is not somethibf tech can replace. Role of tech is improve the jounalism and accessible.
Not big deal
NYT just wants a licensing deal, not a full standoff on whether LLM model output is fair use.
I think the publishers are about to get a juicy deal out of this. The per article fine of the lawsuit would be huge, and a recurring revenue model would be great. Maybe finally this is how the papers can make money again.
This was legal when company was a nonprofit but monetizing the product has caused a lot of lawsuits everywhere because OAI is now profiting from that data and enriching employees. Leadership has no clue how to run a consumer business, theyâre all research PhDs.
They are screwed. They should have been more cautious and thought it through before monetising the product. It is a very expensive mistakes I believe. And not only NYT, tons of other paywalled publishers will do or are doing the same.
This is stupid. First off, how reproducible are these results? OpenAI can tweak the random seed to make reproducibility hard. Second, even if a model spits out some existing content verbatim: they can just enclose the text in quotes and cite where it comes from?
NYT is behind a paywall though. Thatâs the whole point of the lawsuit
Your second point is impossible. The model is not a hash table , it doesnât have any memory of where certain text came from.