Reddit user sparks debate: AI agents feel 'fundamentally wrong'
Cursor moving without your control triggers a deep discomfort, says one user.
A popular Reddit post has struck a nerve in the AI community by articulating a subtle but pervasive discomfort with AI agent demos. The user describes watching a cursor move across the screen, clicking through interfaces, booking flights, and sending emails — all technically impressive but fundamentally unsettling. They realized the unease stems not from distrust in the AI's capabilities, but from the loss of agency: seeing a cursor move on their screen without their physical input feels wrong. This visceral reaction suggests a psychological barrier that many interface designers may be overlooking.
The user contrasts the current 'AI acts' paradigm with a simpler preference: an AI that observes, answers questions like 'what am I looking at?' and advises on next steps — but leaves the action to the human. This raises a critical product question: is the market pushing toward autonomous agents because that's what users truly want, or because it's what showcases best in demos? The debate touches on trust, control, and the fundamental UX of human-AI collaboration. For enterprise tools, this could mean building advisory copilots rather than autonomous agents, especially in high-stakes environments where oversight matters.
- User feels discomfort not from AI capability but from watching an autonomous cursor move without human input.
- Prefers an advisory AI that answers 'what should I do next?' rather than taking over the interface.
- Questions whether the industry's autonomous agent push aligns with actual user preference for assisted control.
Why It Matters
For UX designers and product managers, the post highlights a critical trust barrier in agentic AI adoption.