Refusing to allow continued failures in the AI arena to dampen its enthusiasm for the topic, Apple appears to be working on an AI-powered chatbot for customer support. Aaron Perris, a contributor for MacRumors, spotted code indicating work on a “Support Assistant” for the Apple Support app. The assistant isn’t live yet, but MacRumors has been able to deduce from the code roughly how it will work.
Essentially it looks like Support Assistant will introduce a new layer of chat that customers can access before speaking to an Apple employee. Apple Support currently has a chat feature, but this puts you in touch with a human; the Support Assistant will use generative AI to provide answers to chat questions without getting that human involved, while still allowing you to wait for the human if the chatbot isn’t sorting the problem.
This will of course be setting off all kinds of alarm bells, because Apple’s attempts at AI have thus far been very disappointing: late, endlessly delayed, frequently wrong, and altogether failing to deliver on their promises. There are even alarm bells going off at Apple Park, apparently: the company includes a warning that generative models such as the one Support Assistant is based on may deliver “incorrect, misleading, incomplete, offensive, or harmful outputs,” and that the information provided by Support Assistant is not a substitute for professional advice. Which makes you wonder what it’s for.
There are areas where malfunctioning AI doesn’t really matter. If you use generative AI to create a custom emoji and it looks wrong and creepy, you can delete it and find something else to send your friend. But a tech support bot that gives bad advice (or says offensive things!), regardless of disclaimers, which some will notice but others will filter out as the usual legal small print, is a disaster waiting to happen. Much like an AI doctor, another idea is supposedly making the rounds at Cupertino.
Perhaps Apple will create an AI attorney to deal with the inevitable lawsuits.