Justine Cassell’s research while affiliated with Carnegie Mellon University and other places

Artificial intelligence (AI) technologies have become increasingly integrated into children’s daily lives, influencing learning, social interactions, and creative activities. This chapter provides an overview of key research fields examining children’s learning from, interactions with, and understanding of AI. Current research indicates that AI has the potential to enhance children’s development across multiple domains; however, ethical considerations need to be prioritized. When children engage in learning activities with AI, they may encounter inappropriate, inaccurate, or biased content. Additionally, children’s social interactions with AI may affect their approach to interpersonal interactions. Finally, children’s developing understanding of the world may make them particularly susceptible to attributing human-like properties to AI, undermining their expectations of these technologies. This chapter highlights the importance of future studies focusing on a child-centered design approach, promoting AI literacy, and addressing ethical concerns to fully harness AI’s potential in child development. Recommendations for parents, technology developers, and policymakers are also provided.

This protocol paper outlines an innovative multimodal and multilevel approach to studying the emergence and evolution of how children build social bonds with their peers, and its potential application to improving social artificial intelligence (AI). We detail a unique hyperscanning experimental framework utilizing functional near-infrared spectroscopy (fNIRS) to observe inter-brain synchrony in child dyads during collaborative tasks and social interactions. Our proposed longitudinal study spans middle childhood, aiming to capture the dynamic development of social connections and cognitive engagement in naturalistic settings. To do so we bring together four kinds of data: the multimodal conversational behaviors that dyads of children engage in, evidence of their state of interpersonal rapport, collaborative performance on educational tasks, and inter-brain synchrony. Preliminary pilot data provide foundational support for our approach, indicating promising directions for identifying neural patterns associated with productive social interactions. The planned research will explore the neural correlates of social bond formation, informing the creation of a virtual peer learning partner in the field of Social Neuroergonomics. This protocol promises significant contributions to understanding the neural basis of social connectivity in children, while also offering a blueprint for designing empathetic and effective social AI tools, particularly for educational contexts.

This paper explores the application of machine learning techniques to predict where hedging occurs in peer-tutoring interactions. The study uses a naturalistic face-to-face dataset annotated for natural language turns, conversational strategies, tutoring strategies, and nonverbal behaviours. These elements are processed into a vector representation of the previous turns, which serves as input to several machine learning models. Results show that embedding layers, that capture the semantic information of the previous turns, significantly improves the model's performance. Additionally, the study provides insights into the importance of various features, such as interpersonal rapport and nonverbal behaviours, in predicting hedges by using Shapley values for feature explanation. We discover that the eye gaze of both the tutor and the tutee has a significant impact on hedge prediction. We further validate this observation through a follow-up ablation study.

Hedging is a strategy for softening the impact of a statement in conversation. In reducing the strength of an expression, it may help to avoid embarrassment (more technically, ``face threat'') to one's listener. For this reason, it is often found in contexts of instruction, such as tutoring. In this work, we develop a model of hedge generation based on i) fine-tuning state-of-the-art language models trained on human-human tutoring data, followed by ii) reranking to select the candidate that best matches the expected hedging strategy within a candidate pool using a hedge classifier. We apply this method to a natural peer-tutoring corpus containing a significant number of disfluencies, repetitions, and repairs. The results show that generation in this noisy environment is feasible with reranking. By conducting an error analysis for both approaches, we reveal the challenges faced by systems attempting to accomplish both social and task-oriented goals in conversation.

Hedges play an important role in the management of conversational interaction. In peer tutoring, they are notably used by tutors in dyads (pairs of interlocutors) experiencing low rapport to tone down the impact of instructions and negative feedback. Pursuing the objective of building a tutoring agent that manages rapport with students in order to improve learning, we used a multimodal peer-tutoring dataset to construct a computational framework for identifying hedges. We compared approaches relying on pre-trained resources with others that integrate insights from the social science literature. Our best performance involved a hybrid approach that outperforms the existing baseline while being easier to interpret. We employ a model explainability tool to explore the features that characterize hedges in peer-tutoring conversations, and we identify some novel features, and the benefits of such a hybrid model approach.

Socio-conversational systems are dialogue systems, including what are sometimes referred to as chatbots, vocal assistants, social robots, and embodied conversational agents, that are capable of interacting with humans in a way that treats both the specifically social nature of the interaction and the content of a task. The aim of this paper is twofold: 1) to uncover some places where the compartmentalized nature of research conducted around socio-conversational systems creates problems for the field as a whole, and 2) to propose a way to overcome this compartmentalization and thus strengthen the capabilities of socio-conversational systems by defining common challenges. Specifically, we examine research carried out by the signal processing, natural language processing and dialogue, machine/deep learning, social/affective computing and social sciences communities. We focus on three major challenges for the development of effective socio-conversational systems, and describe ways to tackle them.