Microsoft deleted the over-eager office assistant Clippy some 17 years ago, but the vision of a friendly and upbeat AI assistant has clearly found its way out of the bin. The company is overhauling Copilot, a text-based artificial intelligence tool linked to Windows and other software, adding vision, voice and the ability to solve more complex problems — along with a more “encouraging” personality.
“We’re really in this incredible moment of transition,” says Mustafa Suleiman, CEO of Microsoft AI. “AI companions now see what we see, hear what we hear and speak the same language we use to communicate with each other.”
Copilot has been met with a mixed response so far, with some users complaining about lag or lack of clarity in its responses, but Microsoft is betting that the tool could eventually become an integral part of Windows, Office and more. By incorporating OpenAI’s AI algorithms into software used by hundreds of millions of people, the company is also at the forefront of testing AI’s potential to improve productivity in office work. Google, a major rival, is also bringing AI to office applications, including Gmail and Google Docs.
The new Copilot will be able to converse with users in multiple human voices, handling interruptions and pauses naturally. “You can interrupt in the middle of the stream, and you can also actively listen,” Suleiman says. “And that’s kind of the art of great conversation.”
Suleiman adds that Copilot has also been changed to offer more emotional support to users. “He’s on your team, he supports you, he’s your hype man,” he says. Copilot Voice will be available today in English to users in Australia, Canada, New Zealand, the United Kingdom and the United States, with more countries to follow, the company says.
Microsoft’s assistant Clippy, an anthropomorphized paper clip, was best known for appearing when users opened Word with the infamous line: “It looks like you’re writing a letter…” The product was unpopular; Microsoft concluded that this was partly because the program failed to deliver the human intelligence it promised, forgetting user preferences and repeating itself endlessly. Large language models are much better at mimicking human intelligence, but their behavior can still be strange and unpredictable, which may be a factor in Copilot’s popularity.
