Machine learning models with high accuracy on test data can still produce systematic failures, such as harmful biases and safety issues, when deployed in the real world. To detect and mitigate such failures, practitioners run behavioral evaluation of their models, checking model outputs for specific types of inputs. Behavioral evaluation is important but challenging, requiring that practitioners discover real-world patterns and validate systematic failures. We conducted 18 semi-structured interviews with ML practitioners to better understand the challenges of behavioral evaluation and found that it is a collaborative, use-case-first process that is not adequately supported by existing task- and domain-specific tools. Using these findings, we designed zeno, a general-purpose framework for visualizing and testing AI systems across diverse use cases. In four case studies with participants using zeno on real-world models, we found that practitioners were able to reproduce previous manual analyses and discover new systematic failures.

People work with AI systems to improve their decision making, but often under- or over-rely on AI predictions and perform worse than they would have unassisted. To help people appropriately rely on AI aids, we propose showing them behavior descriptions, details of how AI systems perform on subgroups of instances. We tested the efficacy of behavior descriptions through user studies with 225 participants in three distinct domains: fake review detection, satellite image classification, and bird classification. We found that behavior descriptions can increase human-AI accuracy through two mechanisms: helping people identify AI failures and increasing people's reliance on the AI when it is more accurate. These findings highlight the importance of people's mental models in human-AI collaboration and show that informing people of high-level AI behaviors can significantly improve AI-assisted decision making.

Data scientists require rich mental models of how AI systems behave to effectively train, debug, and work with them. Despite the prevalence of AI analysis tools, there is no general theory describing how people make sense of what their models have learned. We frame this process as a form of sensemaking and derive a framework describing how data scientists develop mental models of AI behavior. To evaluate the framework, we show how existing AI analysis tools fit into this sensemaking process and use it to design AIFinnity , a system for analyzing image-and-text models. Lastly, we explored how data scientists use a tool developed with the framework through a think-aloud study with 10 data scientists tasked with using AIFinnity to pick an image captioning model. We found that AIFinnity ’s sensemaking workflow reflected participants’ mental processes and enabled them to discover and validate diverse AI behaviors.

Interfaces for machine learning (ML), information and visualizations about models or data, can help practitioners build robust and responsible ML systems. Despite their benefits, recent studies of ML teams and our interviews with practitioners (n=9) showed that ML interfaces have limited adoption in practice. While existing ML interfaces are effective for specific tasks, they are not designed to be reused, explored, and shared by multiple stakeholders in cross-functional teams. To enable analysis and communication between different ML practitioners, we designed and implemented Symphony, a framework for composing interactive ML interfaces with task-specific, data-driven components that can be used across platforms such as computational notebooks and web dashboards. We developed Symphony through participatory design sessions with 10 teams (n=31), and discuss our findings from deploying Symphony to 3 production ML projects at Apple. Symphony helped ML practitioners discover previously unknown issues like data duplicates and blind spots in models while enabling them to share insights with other stakeholders.

AI systems can fail to learn important behaviors, leading to real-world issues like safety concerns and biases. Discovering these systematic failures often requires significant developer attention, from hypothesizing potential edge cases to collecting evidence and validating patterns. To scale and streamline this process, we introduce crowdsourced failure reports, end-user descriptions of how or why a model failed, and show how developers can use them to detect AI errors. We also design and implement Deblinder, a visual analytics system for synthesizing failure reports that developers can use to discover and validate systematic failures. In semi-structured interviews and think-aloud studies with 10 AI practitioners, we explore the affordances of the Deblinder system and the applicability of failure reports in real-world settings. Lastly, we show how collecting additional data from the groups identified by developers can improve model performance.

Most of the work on interpretable machine learning has focused on designing either inherently interpretable models, which typically trade-off accuracy for interpretability, or post-hoc explanation systems, whose explanation quality can be unpredictable. Our method, ExpO, is a hybridization of these approaches that regularizes a model for explanation quality at training time. Importantly, these regularizers are differentiable, model agnostic, and require no domain knowledge to define. We demonstrate that post-hoc explanations for ExpO-regularized models have better explanation quality, as measured by the common fidelity and stability metrics. We verify that improving these metrics leads to significantly more useful explanations with a user study on a realistic task.

Ensuring effective public understanding of algorithmic decisions that are powered by machine learning techniques has become an urgent task with the increasing deployment of AI systems into our society. In this work, we present a concrete step toward this goal by redesigning confusion matrices for binary classification to support non-experts in understanding the performance of machine learning models. Through interviews (n=7) and a survey (n=102), we mapped out two major sets of challenges lay people have in understanding standard confusion matrices: the general terminologies and the matrix design. We further identified three sub-challenges regarding the matrix design, namely, confusion about the direction of reading the data, layered relations and quantities involved. We then conducted an online experiment with 483 participants to evaluate how effective a series of alternative representations target each of those challenges in the context of an algorithm for making recidivism predictions. We developed three levels of questions to evaluate users' objective understanding. We assessed the effectiveness of our alternatives for accuracy in answering those questions, completion time, and subjective understanding. Our results suggest that (1) only by contextualizing terminologies can we significantly improve users' understanding and (2) flow charts, which help point out the direction of reading the data, were most useful in improving objective understanding. Our findings set the stage for developing more intuitive and generally understandable representations of the performance of machine learning models.