Microsoft has outlined its belief in users actively collaborating with AI, stating that generative AI's responses can be "usefully wrong".
In a blog about Microsoft's philosophy on AI, the tech giant suggested that generative AI should be considered a creative collaborator rather than a basic answer-providing machine. Microsoft noted that if a user were to ask its AI-powered productivity tool Copilot for 20 ideas to tackle a work challenge, the answers offered would be imperfect but valuable, just as if they were suggestions by a colleague or friend in reality.
"At Microsoft, we like to say that this is when AI is 'usefully wrong'," the company wrote.
It’s a different, more collaborative way of working with technology—one that upends our long-held assumptions about what computers can and can’t do. We’re used to one interaction with a computer: put in a query, get an answer. With AI, the answer isn’t the final word; the magic is in the conversation, the back and forth. And it’s up to people to build on, combine, or transform the content into something original and meaningful."
Microsoft implied that AI's "usefully wrong" answer could offer unexpected or unorthodox ideas that challenge traditional thinking. This is perhaps where Microsoft was keen to distinguish between an odd and original suggestion and an AI "hallucination", which involves misinformation or lies that could problematise users' leveraging of those suggestions.
"AI can offer up content that is unexpected, surprising, or just plain weird," the blog said. "Don't let that discourage you. Instead, embrace the challenge of approaching a problem from a different perspective."
An additional aspect of Microsoft's philosophy on AI is its assertion that it should be used to improve human workflows rather than replace them, and it highlighted vigilance in checking every answer or suggestion it provides: "AI isn't a replacement for your creativity or judgment; it's a tool that can enhance and augment those innately human skills."
Microsoft suggested that users consider relevance, appropriateness, and personality in AI answers and asked users to evaluate how they might improve AI answers constantly — illustrating the company's concept of AI as a collaborative tool that demands user refining, regardless of how sophisticated AI's models are and will eventually become.
