Human-AI collaboration is not very collaborative yet: A taxonomy of interaction patterns in AI-assisted decision making from a systematic review

Current AI systems prioritize technology over human needs: The paper finds that AI development has "disproportionately focused on technological advancements, often overlooking the alignment between algorithmic outputs and human expectations," suggesting that many AI tools are built without adequately considering what users actually need.

Most AI-human interactions are overly simplistic: Across 105 reviewed studies, the researchers discovered that "current interactions are dominated by simplistic collaboration paradigms, leading to little support for truly interactive functionality"—meaning most AI systems today don't enable genuine back-and-forth dialogue between humans and AI.

How AI presents information matters as much as what it says: The paper emphasizes that "how the information is presented, e.g., the sequence of recommendations and solicitation of interpretations, is equally crucial" as providing explanations, highlighting that interaction design choices significantly impact how well humans and AI actually work together.

There's no common language for designing human-AI interactions: The researchers note "a common vocabulary for human-AI interaction protocols is lacking," which means organizations building AI systems lack shared standards or frameworks for thinking about how humans should interact with their tools.

A new taxonomy can guide better AI design: The authors introduce a classification system of interaction patterns to help designers "foster interaction designs for achieving clear communication," offering a practical tool to move beyond current limitations toward more collaborative AI systems.