Apple Tests AI 'Ask' Tool To Enhance Support Services

Apple is reportedly testing a new artificial intelligence (AI) tool dubbed "Ask" specifically designed for AppleCare support advisors. This tool, designed to automate responses to technical queries, leverages data from an internal database to generate solutions. This move is part of Apple's broader exploration into AI technologies, a venture its CEO has expressed enthusiasm about for the upcoming developments in 2024. Though official announcements are pending, the tech giant is believed to be experimenting with various tools that utilize generative models akin to ChatGPT's functionality.

Details have surfaced through a report by MacRumors, revealing the initiation of a pilot programme that equips selected AppleCare support advisors with the "Ask" tool. This programme distinguishes itself from conventional search tools by generating tailored responses based on the specifics of a query, such as the type of device or operating system in question. Advisors interacting with this tool have the option to rate the generated answers as either "helpful" or "unhelpful," enabling a feedback loop for improvement.

The accuracy and reliability of generative language models are critical, given their propensity for "hallucinations" or the generation of confident yet fabricated responses. To mitigate this, the "Ask" tool is trained exclusively on Apple's robust internal database, with additional safeguards to ensure the provided answers are "factual, traceable, and useful." Plans are in place to extend the availability of this tool to more advisors following an initial phase of feedback collection.

The origins of the "Ask" tool might trace back to or be associated with the previously leaked "Ajax," internally dubbed by some as "AppleGPT." Although the specific terminology Apple uses to refer to the "Ask" tool remains uncertain, the language used in the leaked information suggests it falls within the realm of AI. This speculation is supported by the use of terms such as "generate" and "factual," which typically would not apply to a standard search and retrieval system.

Apple's engagement with machine learning (ML) and AI is not a new venture. For years, the tech giant's touch keyboard has utilized an algorithm to predict user input, which has evolved to incorporate a transformer language model for autocorrect functionality. As the tech community buzzes with rumours about Apple's AI ambitions and its potential implications for iOS 18, anticipation grows for the upcoming Worldwide Developers Conference (WWDC) in June, where more details are expected to emerge.