Apple has reportedly ramped up work on its own Large Language Models (LLMs) and other AI-related tools to take on offerings from its competitors such as Google Bard and OpenAI ChatGPT. The company plans to reveal a “significant” AI-powered product as early as next year, according to a new extensive report from Bloomberg‘s Mark Gurman.
---------------
Hey, we know you love your Mac and want to take great care of it. That's why CleanMyMac X has been optimizing Macs for over 15 years — it knows exactly where the clutter hides and understands what every Mac needs. The app is available on the Mac App Store with an average rating of 4,6 out of 5. Give Your Mac Some Love
---------------
Gurman warns that the company is “yet to devise a clear strategy” to release its in-house LLM-powered chatbot to the public as it irons out privacy concerns associated with such services. Nevertheless, Apple has been “quietly working” on LLM-powered frameworks internally dubbed as Ajax. The company is also internally testing a ChatGPT-like chatbot that has been referred to as “Apple GPT” by engineers working on the project, adds Gurman.
The iPhone maker has built its own framework to create large language models — the AI-based systems at the heart of new offerings like ChatGPT and Google’s Bard — according to people with knowledge of the efforts.
Mark Gurman for Bloomberg
The AI efforts are reportedly co-led by Senior Vice President of Machine Learning and AI Strategy John Giannandrea and Senior Vice President of Software Engineering Craig Federighi. Giannandrea prefers a more careful approach to AI development, adds Gurman.
John Giannandrea, the company’s head of machine learning and AI, and Craig Federighi, Apple’s top software engineering executive, are leading the efforts. But they haven’t presented a unified front within Apple, said the people. Giannandrea has signaled that he wants to take a more conservative approach, with a desire to see how recent developments from others evolve.
Mark Gurman for Bloomberg
Apple apparently began its internal AI push several months ago with “several teams collaborating” on the initiative. The company is apparently greatly concerned with missing out on the latest technology trend, hence its frantic approach to catch up to ChatGPT and other chatbots. Gurman reports that work on the Ajax framework – which forms the basis of Apple GPT – began last year as a way for Apple to streamline machine-learning development inside the company.
Ajax is already powering several consumer-facing products such as Siri, Spotlight Search, and Maps, adds Gurman. Apple recently began using Ajax to create LLMs as the “foundation” for Apple GPT. However, the internal rollout of Apple GPT was scrutinized over “security concerns” after it was created towards the end of last year by a small engineering group within Apple. Not all employees are able to use Apple GPT, as access requires special prior approval, adds Gurman.
The chatbot app was created as an experiment at the end of last year by a tiny engineering team. Its rollout within Apple was initially halted over security concerns about generative AI, but has since been extended to more employees. Still, the system requires special approval for access.
Mark Gurman for Bloomberg
Apple employees with access to the chatbot are already using it to help with product prototyping. Like Bard and ChatGPT, Apple’s chatbot can also summarize and answer questions set forward by the user, with employees describing Apple GPT as a replica of the aforementioned services.
Despite the ramp-up of AI efforts inside Apple, the company is reportedly unsure of how to release consumer-facing products powered by generative AI, warns Gurman. Multiple groups are leading work on the company’s AI work including AI engineering, software engineering, and cloud services engineering.
Apple apparently considered signing a cooperation contract with OpenAI to license its technologies after it took a “corporate trial” of the former’s technology, adds Gurman. Apple shelved those plans before shifting its focus to developing Ajax, which is reportedly based on Google’s JAX machine-learning framework.
In April, it was revealed that Apple is working on an AI-based personal coach service that it hopes to launch sometime in 2024. The company was also said to be going on a hiring spree for generative AI engineers across many of its teams, with as many as 28 job openings in May alone.