The latest buzz from the tech world suggests that Apple is gearing up to create its very own Large Language Model (LLM). According to a report from the New York Times, the company is in need of substantial training data for this ambitious project.
To ensure top-notch quality while respecting copyrights, Apple is planning to strike multi-year deals with publishers, looking at a staggering minimum value of $50 million. The goal? To tap into the extensive archives of news content held by these publishers, using it to train and refine Apple’s AI models.
They’ve reached out to big names in the publishing industry like Condé Nast, NBC News, and IAC/InterActiveCorp. But, word on the street is that some publishers are a bit hesitant about Apple’s proposal.
Should these licensing agreements fall into place, Apple might just gain access to the archives of major publications such as Vogue, The New Yorker, The Daily Beast, and People.
This move indicates a potential shift in Apple’s AI strategy. While competitors like Google and Microsoft have been making strides in AI, Apple has been somewhat reserved. However, by making moves to license content from publishers, it seems they’re ready to step up their game in this arena.
Rumor has it that Apple already boasts an in-house AI chatbot named Apple GPT, resembling the functionality of ChatGPT. And they’ve recently introduced AI enhancements to improve autocorrect in iOS 17. There’s a lot of speculation about even more advanced AI features in the pipeline.
By tapping into these news archives, Apple aims to train its AI models using real-world data. This could potentially result in more intelligent and natural AI conversations, addressing a common issue faced by other similar models – sourcing training data from web crawlers without proper licensing. If Apple can ensure that its LLM contains content with impeccable copyright adherence, it could give them a significant edge.
Apple seems to be following the footsteps of other major AI players. For instance, OpenAI has already made deals with Associated Press and Axel Springer, leveraging their news content for AI training purposes.
