(The Atlantic) – Whenever AI companies release new models, die-hard users tend to greet the updates with a wave of grumbling about changes to how a model responds to prompts, how it integrates into their workflow, or, simply, the way it speaks. Most people soon get used to the latest release, ready to gripe again when the next one comes along. But for one niche group of particularly fervent AI users, model updates represent a much more alarming proposition.

These are people who have developed what they consider to be intimate relationships with AI products. For them, AI has become a companion—a kind of friend, partner, or even lover—and any change to the underlying model poses something akin to an existential threat. (Read More)