Reproduce “Human-AI collaboration is not very collaborative yet: A taxonomy of interaction patterns in AI-assisted decision making from a systematic review” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Catalina Gomez, Sue Min Cho, Shichang Ke, Chien-Ming Huang, Mathias Unberath
Researchers at Johns Hopkins University systematically reviewed 105 empirical studies on AI-assisted decision-making to develop a taxonomy of human-AI interaction patterns. The resulting seven-pattern classification reveals that existing human-AI collaboration largely employs simplistic assistance models, indicating a need for more reciprocal and dynamic engagement.
Key insights
- Current human-AI collaboration in decision support is predominantly rudimentary, characterized by 'AI-first' and 'AI-follow' assistance rather than truly interactive or reciprocal engagement.
- The prevalence of simplistic interaction patterns, like AI-first assistance, can introduce cognitive biases such as anchoring and confirmation bias in human decision-making.
- The proposed taxonomy offers a structured language for researchers and designers to deliberately choose or combine interaction patterns, fostering improved communication, trustworthiness, and user agency.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.