Apple is developing its own generative AI technology... but refuses to allow his employees to use ChatGPT or a similar chatbot. The company fears that confidential data may be intercepted by competitors.
Apple has ambitions to enter the world of artificial intelligence. According to the Wall Street Journal, the California giant is currently working on "proprietary technology similar" to ChatGPT. The publication explains that it has studied Apple's internal documents and interviewed "people familiar with this issue."
Linguistic models like OpenAI's GPT, Google's Palm, or Meta's LLaMA are obviously being developed in Apple labs. They are supervised by John Giannandrea, Apple's senior vice president of machine learning and AI. To develop these models, the company can also rely on the innovations of many AI startups that it has acquired in recent years. For example, in 2020, Apple bought Inductiv, which created an AI capable of automating tasks to identify and correct errors in data. Among other acquisitions - Xnor.ai , AI Music and, most recently, WaveOne.
This is not the company's only AI-related project. In parallel, Apple is working on artificial intelligence designed to help owners of Apple Watch. With a paid subscription, this virtual trainer will give users health tips based on their data. Apple fears ChatGPT While preparing its own solution, Apple has explicitly banned its employees from using generative AI to help with work. In particular, the company asked its employees not to use Copilot, an AI from Microsoft that helps developers.
Copilot is capable of generating and completing computer code in several programming languages, such as JavaScript and Python. To get the code, developers must first provide the AI with some of their developments. Apparently, generative AI was regularly used by Apple employees. This restriction is motivated by Apple's concerns about privacy. The giant from Cupertino fears that the data provided by its employees may end up on OpenAI and Microsoft servers. The ChatGPT startup collects all the dialogs of its users by default in order to improve its language models. Like Apple, Samsung has banned its employees from using intelligent chatbots at work. The South Korean giant even threatens to fire all recalcitrant employees, while emphasizing that an alternative intended only for the company is currently being developed. The more moderate giant Amazon has warned its employees about the privacy risks inherent in generative AI.
Like Samsung and Amazon, Apple fears that confidential information about its operating systems or future confidential projects may end up in the hands of competitors or intruders. Note that Apple is known for imposing a real cult of secrecy on its employees. The company has always done everything possible to avoid information leaks before the release of its products. Apple recently caught an employee who was leaking information about its new products to social networks. In this context, banning ChatGPT and others is not a surprise. OpenAI and Microsoft are seeking to reassure companies Aware of the concerns raised by its chatbot, OpenAI plans to launch a subscription for businesses, ChatGPT Business.
In this case, user data will not be used to train the language model. Microsoft, which is the main partner of OpenAI, is also working on a solution to appease companies wishing to use AI. The publisher is developing a more privacy-friendly version of ChatGPT.
The offer guarantees professionals that their data will be stored on dedicated servers, well isolated from the information of other users.