> There’s something about talking to an AI that just always feels wrong.

That the potential for scams and emotional manipulation seem much higher than any "positive" use cases