Guidelines for Human-AI Interaction
By Saleema Amershi, Daniel S. Weld, Mihaela Vorvoreanu, Adam Fourney, Besmira Nushi, Penny Collisson, Jina Suh, Shamsi T. Iqbal, Paul N. Bennett, K. Quinn and 3 more. 2019
2,339 citations recorded by the source corpus.
Advances in artificial intelligence (AI) frame opportunities and challenges for user interface design. Principles for human-AI interaction have been discussed in the human-computer interaction community for over two decades, but more study and innovation are needed in light of advances in AI and the growing uses of AI technologies in human-facing applications. We propose 18 generally applicable design guidelines for human-AI interaction. These guidelines are validated through multiple rounds of evaluation including a user study with 49 design practitioners who tested the guidelines against 20 popular AI-infused products. The results verify the relevance of the guidelines over a spectrum of interaction scenarios and reveal gaps in our knowledge, highlighting opportunities for further research. Based on the evaluations, we believe the set of design guidelines can serve as a resource to practitioners working on the design of applications and features that harness AI technologies, and to researchers interested in the further development of human-AI interaction design principles.
Subfields: Social Psychology, Natural Language Processing, Human-Computer Interaction
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Social Psychology
- M. de Choudhury
- Carolyn Penstein Rosé
- Eric Horvitz
- Michael J. Muller
- Justin D. Weisz
- Morgan Klaus Scheuerman
- Hanspeter Pfister
- Q. Liao
In Natural Language Processing
- Mohit Bansal
- Sebastian Gehrmann
- Percy Liang
- Yonatan Belinkov
- Alexander M. Rush
- Yejin Choi
- Mirella Lapata
- Graham Neubig
In Human-Computer Interaction
- Sebastian Gehrmann
- Hendrik Strobelt
- Q. Liao
- Michael S. Bernstein
- Ece Kamar
- Besmira Nushi
- Eric Horvitz
- Sarah Miller
Citations in the corpus
- AI Chains: Transparent and Controllable Human-AI Interaction by Chaining Large Language Model Prompts (2021, background): “Prior work has noted core challenges of the interaction, such as a lack of transparency, controllability, and user agency [5, 14, 33].”
- Rapid Trust Calibration through Interpretable and Uncertainty-Aware AI (2020, background): “21,28,29 To do this, the AI system must be able to estimate the uncertainty in its outputs.”
- AI-generated characters for supporting personalized learning and well-being (2021, background): “We foresee a future where generative AI will become ubiquitous in our daily lives, as part of the ever growing human–AI interaction landscape, but this is not without challenges 93 .”
- Interactive Model Cards: A Human-Centered Approach to Model Documentation (2022, methodology): “’s guidelines for human-AI interaction [2], as well as by other guidelines and taxonomies for human-centered AI [3, 64] and proposals for enabling human contestation of AI-driven decisions [28, 40].”
- "Happy and Assured that life will be easy 10years from now.": Perceptions of Artificial Intelligence in 8 Countries (2019, neutral): “’s synthesis of twenty years of AI design learnings into 18 guidelines for human-AI interaction design [2], the People + AI Guidebook,7 and other resources outlined in Carney’s summary [9].”
- A Multidisciplinary Survey and Framework for Design and Evaluation of Explainable AI Systems (2018, methodology): “The benefits of interactive system design have been previously explored in the topic of interactive machine learning [6, 7] for novice end-users.”
- “It’s Complicated”: Negotiating Accessibility and (Mis)Representation in Image Descriptions of Race, Gender, and Disability (2021, background): “mending more representative and transparent data, intentionally designing for outliers, and methods which minimize data sharing [3, 34, 35, 44].”
- CodeAid: Evaluating a Classroom Deployment of an LLM-based Programming Assistant that Balances Student and Educator Needs (2024, background): “Recent studies highlight the challenges in user interactions with generative AI models due to their low transparency and predictability [2].”
- The Participatory Turn in AI Design: Theoretical Foundations and the Current State of Practice (2023, background): “…that participation can enable AI systems to better reflect the values, preferences, and needs of users and other impacted stakeholders, or more broadly, that participation will empower stakeholders in shaping the design of AI systems [3, 7, 19, 26, 33, 51, 61, 118, 137, 143, 172, 189, 198–200].”
- Explanation-Based Human Debugging of NLP Models: A Survey (2021, background): “Also, to move towards deployment, it is important to follow human-AI interaction guidelines (Amershi et al., 2019) and evaluate EBHD with potential end users, not just via simulation or crowdsourcing, since human factors play an important role in real situations (Amershi et al.”
- Human-Centered Artificial Intelligence: Reliable, Safe & Trustworthy (2020, neutral): “…& Kristensson, 2018) paved the way for Microsoft’s Guidelines for AI-Human Interaction (https://www.microsoft.com/en-us/research/project/guidelines-for-human-ai-interaction/), which has 18 guidelines for initial use, normal use, coping with problems, and changes over time (Amershi et al., 2019).”
- How do Data Science Workers Collaborate? Roles, Workflows, and Tools (2020, background): “Based on the empirical findings and design suggestions from previous literature [11, 32, 43, 61, 67, 75, 83, 91], some designers and system builders have proposed human-in-the-loop design principles for science tools [2, 3, 28, 49, 50, 92].”
- Trustworthy AI: From Principles to Practices (2021, background): “As an extension of human–computer interaction (HCI), human–AI interaction has aroused wide attention in the AI industry [4, 18].”
- Generative Agents: Interactive Simulacra of Human Behavior (2023, background): “However, in other application domains, it will be important to follow best practices in human-AI design [5, 107] to understand errors and how they might percolate into the user experience.”
- How Platform-User Power Relations Shape Algorithmic Accountability: A Case Study of Instant Loan Platforms and Financially Stressed Users in India (2022, background): “As an immediate call-to-action, we urge designers to implement AI systems based on established industry practices [15, 96].”
- "Hello AI": Uncovering the Onboarding Needs of Medical Practitioners for Human-AI Collaborative Decision-Making (2019, background): “In doing so, it also provides a clinical angle to recent human–AI interaction guidelines, which state that AI systems should make their capabilities clear to users upon initial interaction with the system [5].”
- Re-examining Whether, Why, and How Human-AI Interaction Is Uniquely Difficult to Design (2020, methodology): “Prior research has been addressing these challenges by promoting interaction design guidelines [4, 35].”