# Human–AI Interaction

How people perceive, trust, and collaborate with AI systems. This sub-article expands [Section 2.2 of Philosophy](../philosophy.md).

## Trust and Calibration
- People over- or under-trust algorithmic advice.
- Calibration requires transparency, feedback, and appropriate reliance.

## Anthropomorphism
- Users attribute intent, emotion, and personality to chatbots (the ELIZA effect).
- This can increase engagement but also create unrealistic expectations.

## Bias and Fairness
- Models reflect and amplify human biases in hiring, lending, policing.
- Fairness research studies disparate impact and mitigation.

## Overreliance and Errors
- Users may defer to AI even when it errs ("automation bias").
- Design must support questioning and verification.

## Why It Matters
Interaction design determines whether AI helps or harms, and how people understand machine behavior.

## Sources
- Weizenbaum, E. (1966). *ELIZA.*
- Bansal, G. et al. (2021). *Does the Whole Exceed its Parts?* (human-AI teams).

*Back to [Philosophy](../philosophy.md).*