Microsoft / Elyse Betters Picaro / ZDNET
Microsoft's on a relentless quest to embed its Copilot AI into every software product it owns. That campaign takes another big step forward today with the addition of some new generative AI features in the company's Edge browser.
Today's release beefs up the capabilities of Copilot Mode, the chat-based search interface for Edge, on Windows PCs and Macs. You can use natural voice commands to navigate pages, and the chatbot can see all the open tabs in the current browser window instead of being limited to whatever's on the current page.
Also: Microsoft gives Copilot a face - here's how to try your new Appearance chat buddy
If that sounds appealing, you can go to the Copilot Mode sign-in page and flip a switch that lets you enable Copilot Mode in Edge Settings. That replaces the old search box with a Copilot chat input box that's designed to handle search, chat, and web navigation. It also gives you the option to open Copilot Mode in a sidebar, where you can ask it questions about the current set of open tabs.
The new tab page for Copilot Mode combines chat and search. Image credit: Microsoft
For now, at least, this feature requires users to opt in. It's also free, but only for a limited time, Microsoft says, without offering any clues as to what it will cost and when the meter will start running.
Also: Microsoft's Copilot Vision can now see and analyze your entire Windows desktop
The blog post announcing today's changes is typically upbeat, anthropomorphizing the Copilot features to put a friendly face on what is, deep down, a computational exercise happening in Microsoft's data centers. It also makes some pretty wild promises along the way: "Copilot understands your intent and helps you get started faster," we are told. "Copilot Mode also sees the full picture across your open tabs, and you can even instruct it to handle some tasks."
Well, no, neither of those statements is true. The Copilot generative AI tool is not human. It is not sentient. It cannot possibly "understand your intent" or "see the full picture" of anything. What it can do is make inferences about those open tabs based on its training data and whatever history it's saved from its interactions with you, generate some text from those inferences in response to your prompt, and then present the results to you in a way that is diabolically persuasive. There's a very good chance it will get some of those things right. But it might also offer up some outright fabrications, as the disclaimers keep warning you:
... continue reading