In a 2005 discussion within the Apple community, shortly after the release of the iPod Shuffle, many users expressed dissatisfaction that the device didn't shuffle songs adequately, often playing tracks from the recent past and lacking randomness. As a result, Apple had to adopt a more effective shuffling algorithm.

I foresee a similar situation arising with AI, where users might argue that its outputs are too flawless and request that it occasionally makes mistakes or displays more human-like qualities, as it may not feel sufficiently human. 

You can find the original discussion here: https://discussions.apple.com/thread/250637.