skip to main content

This content will become publicly available on March 22, 2023

Title: Do People Engage Cognitively with AI? Impact of AI Assistance on Incidental Learning
When people receive advice while making difficult decisions, they often make better decisions in the moment and also increase their knowledge in the process. However, such incidental learning can only occur when people cognitively engage with the information they receive and process this information thoughtfully. How do people process the information and advice they receive from AI, and do they engage with it deeply enough to enable learning? To answer these questions, we conducted three experiments in which individuals were asked to make nutritional decisions and received simulated AI recommendations and explanations. In the first experiment, we found that when people were presented with both a recommendation and an explanation before making their choice, they made better decisions than they did when they received no such help, but they did not learn. In the second experiment, participants first made their own choice, and only then saw a recommendation and an explanation from AI; this condition also resulted in improved decisions, but no learning. However, in our third experiment, participants were presented with just an AI explanation but no recommendation and had to arrive at their own decision. This condition led to both more accurate decisions and learning gains. We hypothesize more » that learning gains in this condition were due to deeper engagement with explanations needed to arrive at the decisions. This work provides some of the most direct evidence to date that it may not be sufficient to provide people with AI-generated recommendations and explanations to ensure that people engage carefully with the AI-provided information. This work also presents one technique that enables incidental learning and, by implication, can help people process AI recommendations and explanations more carefully. « less
Authors:
;
Award ID(s):
2107391
Publication Date:
NSF-PAR ID:
10366302
Journal Name:
27th International Conference on Intelligent User Interfaces (IUI ’22)
Page Range or eLocation-ID:
794 to 806
Sponsoring Org:
National Science Foundation
More Like this
  1. In this paper, we consider the problem of learning Boolean formulae from examples obtained by actively querying an oracle that can label these examples as either positive or negative. This problem has received attention in both machine learning as well as formal methods communities, and it has been shown to have exponential worst-case complexity in the general case as well as for many restrictions. In this paper, we focus on learning sparse Boolean formulae which depend on only a small (but unknown) subset of the overall vocabulary of atomic propositions. We propose two algorithms—first, based on binary search in the Hamming space, and the second, based on random walk on the Boolean hypercube, to learn these sparse Boolean formulae with a given confidence. This assumption of sparsity is motivated by the problem of mining explanations for decisions made by artificially intelligent (AI) algorithms, where the explanation of individual decisions may depend on a small but unknown subset of all the inputs to the algorithm. We demonstrate the use of these algorithms in automatically generating explanations of these decisions. These explanations will make intelligent systems more understandable and accountable to human users, facilitate easier audits and provide diagnostic information in themore »case of failure. The proposed approach treats the AI algorithm as a black-box oracle; hence, it is broadly applicable and agnostic to the specific AI algorithm. We show that the number of examples needed for both proposed algorithms only grows logarithmically with the size of the vocabulary of atomic propositions. We illustrate the practical effectiveness of our approach on a diverse set of case studies.« less
  2. Product retrieval systems have served as the main entry for customers to discover and purchase products online. With increasing concerns on the transparency and accountability of AI systems, studies on explainable information retrieval has received more and more attention in the research community. Interestingly, in the domain of e-commerce, despite the extensive studies on explainable product recommendation, the studies of explainable product search is still in an early stage. In this paper, we study how to construct effective explainable product search by comparing model-agnostic explanation paradigms with model-intrinsic paradigms and analyzing the important factors that determine the performance of product search explanations. We propose an explainable product search model with model-intrinsic interpretability and conduct crowdsourcing to compare it with the state-of-the-art explainable product search model with model-agnostic interpretability. We observe that both paradigms have their own advantages and the effectiveness of search explanations on different properties are affected by different factors. For example, explanation fidelity is more important for user's overall satisfaction on the system while explanation novelty may be more useful in attracting user purchases. These findings could have important implications for the future studies and design of explainable product search engines.
  3. Background/Context: After-school programs that focus on integrating computer programming and mathematics in authentic environments are seldomly accessible to students from culturally and linguistically diverse backgrounds, particularly bilingual Latina students in rural contexts. Providing a context that broadens Latina students’ participation in mathematics and computer programming requires educators to carefully examine how verbal and nonverbal language is used to interact and to position students as they learn new concepts in middle school. This is also an important stage for adolescents because they are likely to make decisions about their future careers in STEM. Having access to discourse and teaching practices that invite students to participate in mathematics and computer programming affords them opportunities to engage with these fields. Purpose/Focus of Study: This case study analyzes how small-group interactions mediated the positionings of Cindy, a bilingual Latina, as she learned binary numbers in an after-school program that integrated computer programming and mathematics (CPM). Setting: The Advancing Out-of-School Learning in Mathematics and Engineering (AOLME) program was held in a rural bilingual (Spanish and English) middle school in the Southwest. The after-school program was designed to provide experiences for primarily Latinx students to learn how to integrate mathematics with computer programming using Raspberry Pimore »and Python as a platform. Our case study explores how Cindy was positioned as she interacted with two undergraduate engineering students who served as facilitators while learning binary numbers with a group of three middle school students. Research Design: This single intrinsic case focused on exploring how small-group interactions among four students mediated Cindy’s positionings as she learned binary numbers through her participation in AOLME. Data sources included twelve 90-minute video sessions and Cindy’s journal and curriculum binder. Video logs were created, and transcripts were coded to describe verbal and nonverbal interactions among the facilitators and Cindy. Analysis of select episodes was conducted using systemic functional linguistics (SFL), specifically language modality, to identify how positioning took place. These episodes and positioning analysis describe how Cindy, with others, navigated the process of learning binary numbers under the stereotype that female students are not as good at mathematics as male students. Findings: From our analysis, three themes that emerged from the data portray Cindy’s experiences learning binary numbers. The major themes are: (1) Cindy’s struggle to reveal her understanding of binary numbers in a competitive context, (2) Cindy’s use of “fake it until you make it” to hide her cognitive dissonance, and (3) the use of Spanish and peers’ support to resolve Cindy’s understanding of binary numbers. The positioning patterns observed help us learn how, when Cindy’s bilingualism was viewed and promoted as an asset, this social context worked as a generative axis that addressed the challenges of learning binary numbers. The contrasting episodes highlight the facilitators’ productive teaching strategies and relations that nurtured Cindy’s social and intellectual participation in CPM. Conclusions/Recommendations: Cindy’s case demonstrates how the facilitator’s teaching, and participants’ interactions and discourse practices contributed to her qualitatively different positionings while she learned binary numbers, and how she persevered in this process. Analysis of communication acts supported our understanding of how Cindy’s positionings underpinned the discourse; how the facilitators’ and students’ discourse formed, shaped, or shifted Cindy’s positioning; and how discourse was larger than gender storylines that went beyond classroom interactions. Cindy’s case reveals the danger of placing students in “struggle” instead of a “productive struggle.” The findings illustrated that when Cindy was placed in struggle when confronting responding moves by the facilitator, her “safe” reaction was hiding and avoiding. In contrast, we also learned about the importance of empathetic, nurturing supporting responses that encourage students’ productive struggle to do better. We invite instructors to notice students’ hiding or avoiding and consider Cindy’s case. Furthermore, we recommend that teachers notice their choice of language because this is important in terms of positioning students. We also highlight Cindy’s agency as she chose to take up her friend’s suggestion to “fake it” rather than give up.« less
  4. Explanations can help users of Artificial Intelligent (AI) systems gain a better understanding of the reasoning behind the model’s decision, facilitate their trust in AI, and assist them in making informed decisions. Due to its numerous benefits in improving how users interact and collaborate with AI, this has stirred the AI/ML community towards developing understandable or interpretable models to a larger degree, while design researchers continue to study and research ways to present explanations of these models’ decisions in a coherent form. However, there is still the lack of intentional design effort from the HCI community around these explanation system designs. In this paper, we contribute a framework to support the design and validation of explainable AI systems; one that requires carefully thinking through design decisions at several important decision points. This framework captures key aspects of explanations ranging from target users, to the data, to the AI models in use. We also discuss how we applied our framework to design an explanation interface for trace link prediction of software artifacts.
  5. Category learning is fundamental to cognition, but little is known about how it proceeds in real-world environments when learners do not have instructions to search for category-relevant information, do not make overt category decisions, and do not experience direct feedback. Prior research demonstrates that listeners can acquire task-irrelevant auditory categories incidentally as they engage in primarily visuomotor tasks. The current study examines the factors that support this incidental category learning. Three experiments systematically manipulated the relationship of four novel auditory categories with a consistent visual feature (color or location) that informed a simple behavioral keypress response regarding the visual feature. In both an in-person experiment and two online replications with extensions, incidental auditory category learning occurred reliably when category exemplars consistently aligned with visuomotor demands of the primary task, but not when they were misaligned. The presence of an additional irrelevant visual feature that was uncorrelated with the primary task demands neither enhanced nor harmed incidental learning. By contrast, incidental learning did not occur when auditory categories were aligned consistently with one visual feature, but the motor response in the primary task was aligned with another, category-unaligned visual feature. Moreover, category learning did not reliably occur across passive observation ormore »when participants made a category-nonspecific, generic motor response. These findings show that incidental learning of categories is strongly mediated by the character of coincident behavior.« less