Apple seeks news archives to train its AI models for better services

Apple news AI Apple is reportedly in talks with news publishers to access their archives of news articles for training its artificial intelligence (AI) models, which could improve its products and services such as Siri, Spotlight, and News app.

In a strategic move that underscores Apple’s commitment to advancing its artificial intelligence (AI) capabilities, the tech giant is reportedly engaged in discussions with news publishers to gain access to extensive archives of news articles. This initiative aims to bolster Apple’s AI models, potentially enhancing products and services like Siri, Spotlight, and the News app. Apple news AI

According to reports by Bloomberg, Apple is offering multiyear deals, with a substantial investment of at least $50 million, to license news article archives from prominent companies such as Condé Nast and NBC News. The primary objective is to leverage these archives for training its generative AI systems. Generative AI, a subfield of AI, focuses on creating new content—text, images, or audio—based on existing data.

While competitors like Google and Microsoft have made strides in launching generative AI products, Apple has primarily used AI for refining photos and autocorrecting text. The company has developed its own large language model (LLM) framework named Ajax and is rumored to be working on a chatbot internally known as Apple GPT. These initiatives, however, are yet to be seamlessly integrated into Apple’s consumer-facing products. Apple news AI

Apple envisions integrating AI across various platforms, including Siri, Messages, and Apple Music. For the latter, AI could facilitate the creation of auto-generated playlists, a feature analogous to Spotify’s offerings through its collaboration with OpenAI. Additionally, Apple is exploring AI applications in Xcode to assist app developers in streamlining their workflow.

Leading the charge on Apple’s AI front are key executives, including John Giannandrea (SVP of machine learning and AI strategy), Craig Federighi (SVP of software engineering), and Eddy Cue (head of services). The company is on track to allocate a substantial budget of $1 billion annually for the development of generative AI products.

Despite Apple’s reputation for secrecy, it has contributed to the open-source development of AI tools, such as MLX, a machine learning framework tailored for Apple silicon. The company has also made strides in AI research, publishing research papers and recruiting notable AI researchers.

However, the decision to utilize news archives to train AI models raises ethical and legal considerations. Questions around the quality and bias of news sources, the privacy and consent of news authors and subjects, and the potential misuse or abuse of generated content may arise. Apple, known for its guarded approach, has not officially commented on its AI deals with news publishers.

As Apple endeavors to push the boundaries of AI, this move marks a crucial step in harnessing the wealth of information contained in news archives to shape the future of its AI-powered products and services.

Apple uses AI in its products, but hasn’t launched a generative AI product along the lines of OpenAI’s ChatGPT or Google Bard. Instead, Apple’s AI is used for things like improving photos and autocorrecting text. The company has built its own large language model (LLM) framework, known as Ajax, as well as a rumored chatbot, known internally as Apple GPT, but neither has been incorporated into its products yet.

The company is looking to integrate AI into Siri, Messages, and Apple Music. For the latter, Apple would use AI to create auto-generated playlists like Spotify does via its partnership with OpenAI. The company is also exploring using AI in Xcode to assist app developers.

The Apple executives leading the AI initiatives are John Giannandrea, SVP of machine learning and AI strategy; Craig Federighi, SVP of software engineering; and Eddy Cue, head of services.

The spending on news archives comes as Apple plays catch-up to some competitors who have already debuted new AI products and features, such as Google, Microsoft, and Amazon. Apple is on track to spend $1 billion per year on developing its generative AI products, Bloomberg reported.

Apple is known for its secrecy and control over its products and services, but it has also contributed to the open source development of AI tools, such as MLX, a machine learning framework specifically designed for Apple silicon. The company has also published some of its AI research papers and hired prominent AI researchers.

Apple’s move to use news to train its AI models could raise some ethical and legal issues, such as the quality and bias of the news sources, the privacy and consent of the news authors and subjects, and the potential misuse or abuse of the generated content. Apple has not commented on its AI deals with news publishers.