Apple is reportedly allocating substantial daily investments into artificial intelligence, as disclosed in a recent report by The Information, a tech and IT website. The company is actively engaged in the development of various AI models through multiple teams.
One of Apple’s AI-focused divisions, referred to as “Foundational Models”, boasts approximately 16 members, including former Google engineers. This division is led by John Giannandrea, Apple’s AI chief, who was brought on board in 2018 to enhance Siri. Giannandrea has purportedly expressed doubts to colleagues regarding the practicality of AI language model-driven chatbots.
Furthermore, Apple has other teams dedicated to AI research. A Visual Intelligence unit is actively crafting an image generation model, while another group delves into “multimodal AI,” capable of recognizing and generating images, video, and text.
These AI models are envisioned to serve diverse purposes. Apple is developing a chatbot to engage with customers using AppleCare services, as well as another to facilitate the automation of complex tasks with Siri.
Insiders involved in the development of these AI models reveal that Apple’s most advanced Language Model, internally known as Ajax GPT, has been trained on an impressive “more than 200 billion parametres” and surpasses the capabilities of OpenAI’s GPT-3.5, which underpins the initial version of ChatGPT introduced last year. This model was initially created for internal use and remains tightly controlled within the company.