Apple spend ‘millions of dollars a day’ training conversational AI

SAN FRANCISCO: Apple is reportedly investing millions of dollars a day in creating multiple artificial intelligence (AI) conversational models, as the generative AI race heats up.

According to a report in The Information, the tech giant is working on multiple AI models across several teams.

“One of its goals is to develop features such as one that allows iPhone customers to use simple voice commands to automate tasks involving multiple steps, according to people familiar with the effort,” the report said late on Wednesday.

The technology might allow users to tell the Siri voice assistant on their phone to create a GIF using the last five photos they’ve taken and text it to a friend, the report added.

Currently, an iPhone user has to manually programme the individual actions.

John Giannandrea, Apple’s head of AI, had authorised the formation of a team to develop conversational AI, known as large-language models four years back.

This was long before ChatGPT came into existence.

A chatbot is in the works would “interact with customers who use AppleCare” while another may make it easier to automate multistep tasks with Siri.

Apple’s most advanced LLM, known internally as Ajax GPT, has been trained on “more than 200 billion parameters” and is more powerful than OpenAI’s GPT-3.5, according to the report, citing sources.

The tech giant was yet to comment on the report.

Apple CEO Tim Cook revealed last month that the tech giant has been working on generative AI and other models for years.

Cook said that Apple views AI and machine learning (ML) as fundamental core technologies.

“And they are virtually embedded in every product that we build,” Cook told CNBC.

“On a research basis, we’ve been doing research on AI and machine learning, including generative AI, for years,” Cook was quoted as saying.

Apple does not talk much about generative AI and it was largely missing from its latest quarterly report too.