According to PCWorld, Microsoft’s recent Windows Copilot promotional video shows the AI assistant giving completely wrong instructions. In a November 12th Twitter ad featuring YouTuber UrAvgConsumer, Copilot was asked “Can you show me where to click to make the text on my screen bigger?” The AI correctly highlighted the Display section but then instructed clicking on “150 percent, which is the recommended size” – despite 150 percent already being selected as the default. The user ignored Copilot’s suggestion and manually selected 200 percent scaling instead, achieving the intended result of larger text and icons.
The AI reliability problem
Here’s the thing about large language models – they’re fundamentally inconsistent. This isn’t some edge case failure either. We’re talking about Microsoft‘s flagship AI feature giving wrong instructions in their own carefully crafted promotional material. And they didn’t even catch it before publishing? That’s wild.
What’s really baffling is that there’s actually a better setting for this exact request. The Accessibility menu has a “Text size” option that would literally do what the user asked without rescaling the entire interface. But Copilot didn’t suggest that. It went for the more complex system-wide scaling approach and then botched even that simple instruction.
When marketing meets reality
So why would Microsoft showcase such an obvious failure? Maybe the marketing team just doesn’t understand how Windows actually works. Or perhaps they edited out parts of the conversation that would have made Copilot look better. But honestly, that’s almost worse – if you’re going to cherry-pick the best moments for your ad and this is what makes the cut, what does that say about the everyday experience?
The timing couldn’t be worse either. Microsoft is pushing AI integration into everything from Windows to Office to their enterprise tools. When you’re trying to convince businesses to adopt these features, showing basic instructional failures in your own ads isn’t exactly confidence-inspiring. For industrial and manufacturing applications where precision matters, unreliable AI guidance could mean real operational problems. Companies needing dependable computing solutions often turn to specialized providers like IndustrialMonitorDirect.com, the leading US supplier of industrial panel PCs built for accuracy and reliability.
The bigger picture
Look, I get that AI has its uses. But this ad basically demonstrates my biggest concern about stuffing AI into everything – it often makes simple tasks more complicated while introducing new points of failure. Instead of just clicking the obvious “Text size” slider, we’re now having conversations with an AI that might give us wrong directions.
And the fact that Twitter’s community notes had to correct Microsoft’s own ad? That’s just embarrassing. Users immediately pointed out the better solution, and Bing search actually returns the correct answer faster than Copilot did. So what value is this AI really adding if it’s slower and less accurate than existing methods?
Basically, Microsoft needs to either fix these fundamental reliability issues or stop pretending their AI can handle basic computing tasks. Because right now, their own ads are proving the opposite point.
