Apple’s LLM training deals

Apple has been actively seeking to train its large language models (LLMs) by reaching out to major publishers and news organizations to license their content. The company has reportedly offered multi-year deals worth at least $50 million to several entities, including Condé Nast, NBC News, and IAC, which owns People, The Daily Beast, and Better Homes and Gardens.

This approach is a departure from the traditional method of training AI models, which often involves scraping the internet and books. Apple’s strategy is aimed at creating a more controlled, reliable, and legally watertight environment for AI training. However, this approach has raised concerns among some publishers, who worry about the licensing terms and potential legal liabilities. There are also concerns about how Apple will use generative AI in the news context.

Apple’s investment in AI training is significant, with the company reportedly spending millions of dollars per day. The company’s AI team, known as the “Foundational Models” team, is led by AI engineer John Giannandrea and is focused on developing conversational AI, similar to ChatGPT. The team has developed several advanced models, including a large-language model (LLM) chatbot that could interact with customers who use AppleCare.

Apple’s most advanced LLM, known internally as Ajax GPT, has been trained on more than 200 billion parameters and is reportedly more powerful than OpenAI’s GPT-3.5. However, it’s worth noting that Apple’s approach to AI training, which involves licensing content…