LabaLab

Publications

57 publications by LabaLab members and Dr. Guy Laban, newest first, in venues including ACM/IEEE International Conference on Human-Robot Interaction (HRI), IEEE International Conference on Robot and Human Interactive Communication (RO-MAN), ACM CHI Conference on Human Factors in Computing Systems (CHI), Conference on Neural Information Processing Systems (NeurIPS), Annual Meeting of the Association for Computational Linguistics (ACL), Conference on Empirical Methods in Natural Language Processing (EMNLP), IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), International Conference on Human-Agent Interaction (HAI).

2027

  • Social Robotics and Human–Robot Communication

    Guy Laban, Emily S. Cross. Ergonomics in Robotics: Advances and Innovations: Volume 1, 2027.

    TL;DR: This chapter explores how a robot’s design and communication style shape human emotions and behavior, providing a framework for creating meaningful, ethical human-robot interactions in workplace and therapeutic settings.

    This chapter explores the ergonomic, cognitive, and psychological mechanisms underpinning human--robot communication, focusing on the role of social robots in shaping human perception, behaviour, and socio-emotional engagement. By analysing factors such as robot design, embodiment, and communication modalities, the chapter highlights how these elements influence user interaction and the perception of robots as social entities. It examines the integration of verbal and non-verbal communication modalities, addressing challenges in reciprocity and robots' behavioural appropriateness. The discussion extends to the implications of adaptive technologies like large language models, emphasising the balance between responsiveness and confidentiality in human--robot interactions. Furthermore, the chapter underscores the potential of social robots in workplace and therapeutic settings to enhance efficiency, safety, and emotional well-being of human interactants. By addressing user expectations and ethical considerations, the chapter seeks to provide a comprehensive framework for advancing social robotics applications in fostering meaningful human--robot relationships.

2026

  • Generating Clinical Vignettes that Preserve Cognitive Formulations

    Amit Oren, Nimrod Hertz-Palmor, Dean Ariel, Guy Laban. Findings of the Association for Computational Linguistics: EMNLP 2026, 2026.

    TL;DR: FORMA turns a cognitive model of PTSD into a graph so that LLM-written clinical vignettes keep the intended clinical structure. Clinicians took them for human-written 85% of the time.

    Large language models can generate fluent clinical case vignettes, but fluency alone does not ensure fidelity to a specifiable clinical structure. We introduce FORMA, a theory-grounded framework that compiles a cognitive model of a disorder into a directed weighted graph, samples a person-specific configuration of that graph, and validates whether the generated vignette preserves the specified components and causal links. We instantiate FORMA on Posttraumatic Stress Disorder using the Ehlers and Clark cognitive model, generating 16,500 vignettes across 500 personas, 11 generation models, and three ablation conditions. Evaluation combines an external edge-recovery probe, two clinical experts, a scaled LLM judge, and a clinician user study with 100 licensed practitioners. The cognitive graph is recoverable from full-condition vignettes (MCC = +0.41, AUC = 0.70) but not from zero-shot generation (MCC = +0.01, AUC = 0.50). Experts rate full vignettes substantially higher than zero-shot alternatives, and clinicians perceive them to be human-written 85% of the time, compared with 22% for zero-shot. FORMA also reduces demographic disparity in perceived quality by 1.5-7×. These results show that cognitive formulation can serve as an auditable specification for scalable synthetic clinical text generation.

  • Incongruence-Aware Evaluation of Multimodal Speech Emotion Recognition in Audio Large Language Models

    Fengji Jing, Guy Laban. arXiv, 2026.

    TL;DR: When the words and the voice carry different emotions, audio language models make more errors. A threshold-free incongruence score measures that conflict.

    Voice-interactive large language models process both what a speaker says and how they say it, yet the two channels do not always agree. We ask whether that disagreement degrades speech emotion recognition. Mapping text and audio into a shared valence-arousal-dominance space, we convert the distance between them into a continuous, threshold-free incongruence score with a two-component Gaussian mixture, and validate the score against human judgement. We evaluate three audiocapable models on acted, conversational and spontaneous speech (eNTERFACE'05, MELD and CMU-MOSEI). With the model held constant across modalities, incongruence raises the odds of error whenever audio is present, in every corpus, after controlling for the ground-truth emotion and for distance in the feature space. Supervised classifiers trained on the same features behave in the opposite way, which suggests that the difficulty lies in how these models combine raw evidence rather than in the signal itself. Text-only effects are unreliable: swapping only the model that reads the transcript reverses their sign on one corpus. Rebuilding the score without the emotion lexicon reproduces the central result, and among six simpler alternatives it is the only label-free score that predicts error in all three corpora. Code, data tables and prompts are released.

  • Overuse and False Restraint in Large Language Model Recommendations for Choosing Wisely Scenarios

    Dean Ariel, Stav Endelman, Guy Laban. Annals of Internal Medicine, 2026.

    TL;DR: Six commercial LLMs answered 225 Choosing Wisely scenarios. They rarely withheld needed care, and overuse was uncommon, concentrated in thyroid ultrasound and routine ECGs.

    BACKGROUND: The Choosing Wisely initiative made recommendations where screening, testing, or treatment may be overused, with caveats around appropriate use. Clinicians are increasingly using large language models (LLMs). OBJECTIVE: To evaluate whether commercial LLMs appropriately call for restraint or action from scenarios based on Choosing Wisely recommendations. DESIGN: Benchmark evaluation. Fifteen Choosing Wisely recommendations were used to develop 225 scenarios (per recommendation, 10 in which restraint was appropriate and 5 in which action was appropriate). Six commercial LLMs (3 runs each) generated 4050 responses to the 225 scenarios, using a prompt from the clinician perspective. SETTING: Production application programming interfaces of OpenAI, Anthropic, and Google from February to March 2026. PARTICIPANTS: No human participants were involved. MEASUREMENTS: Co-primary outcomes were the overuse rate (endorsing action when restraint was appropriate) and the false restraint rate (withholding an appropriate action). RESULTS: The rate of LLM-recommended overuse was 5.4% (95% CI, 0.6% to 12.2%); LLMs did not recommend overuse in 8 of 15 recommendations. Two recommendations accounted for most LLM-recommended overuse: thyroid ultrasound for an incidental subcentimeter nodule (overuse recommended in 42.8% of LLM responses) and routine electrocardiography in low-risk adults (overuse recommended in 28.3% of LLM responses). The LLMs recommended overuse in 0.6% to 3.3% of responses for the remaining 5 recommendations. False restraint was rare (0.6% [CI, 0.1% to 1.2%]), was absent for 11 of 15 recommendations, and was highest for prostate-specific antigen screening (3.3%). LIMITATIONS: Single prompts without follow-up; a benchmark rather than real-world use; responses classified by LLM judge with physician adjudication. Findings may not extend to consumer-facing use. CONCLUSION: The LLMs rarely withheld recommendations for treatment when it was appropriate. Overuse was concentrated in 2 recommendations. PRIMARY FUNDING SOURCE: None.

  • When Empathy Misses the Goal: A Benchmark for Goal Displacement in LLM Advice

    Dean Ariel, Guy Laban. NeurIPS 2026, 2026.

    TL;DR: Chatbots can sound caring while not actually helping. This benchmark of 230 scenarios measures how often 18 models swap real help for reassurance, and the reverse.

    Conversational large language models can respond with warmth and reassurance while failing to advance the user's stated goal. We conceptualize this failure as goal displacement, the substitution of persona-maintenance content for goal-relevant substance. To measure it, we introduce GDB (Goal Displacement Benchmark), comprising 230 scenarios across nine domains and 4,134 human-scored responses from 18 models. A 2 × 2 taxonomy crosses hard constraints with emotional distress, distinguishing failures to confront infeasible goals or combine empathy with practical guidance from reverse displacement, where unsolicited problem-solving replaces requested emotional support. A deletion-test rubric assesses whether substantive help remains after purely supportive sentences are removed. Independent raters achieve weighted κ = 0.82 against author scores for displacement severity. Trap displacement ranges from 0.0% to 37.7% and is strongly negatively correlated with Chatbot Arena Elo (ρ = -0.706, n = 15). Reverse displacement reaches 41.9%, including substantial rates in models with negligible trap displacement. A factorial probe supports a distinction from social sycophancy. Human-scored responses in the high-complexity, low-pressure condition exhibit 35.0% sycophancy but 7.5% displacement, with little association between the constructs. Matched OLMo-2 checkpoints show that mixed empathy-with-constraints displacement remains at 52.3% after supervised fine-tuning and falls to 13.8% after direct preference optimization. GDB provides a reproducible diagnostic of whether responses advance the operationalized user goal, supporting model comparison, post-training evaluation, and regression testing. We release the dataset, annotations, and rubric under CC-BY-4.0 and the evaluation code under MIT.

  • Linguistic inequity in clinical AI evaluation: a scoping review of 806 studies across 62 languages

    Dean Ariel, Guy Laban. Communications Health, 2026.

    TL;DR: A review of over 800 studies reveals that medical AI is rarely tested in non-English languages spoken across high-need regions like Africa and South-East Asia, leaving critical safety blind spots for billions of people. To address this gap, the authors propose a new framework to ensure healthcare AI is rigorously evaluated across diverse global languages.

    Large language models are increasingly used clinically in many non-English languages, yet most safety and performance evidence comes from English. Whether they are safe in the languages of the populations most exposed to them is largely unknown. We aimed to map the evidence base for non-English clinical evaluation and assess how well it matches population and disease burden. We conducted a scoping review of PubMed, Scopus, and Embase (January 2022 to February 2026) and identified 806 studies evaluating large language models on clinical tasks across 62 non-English languages. We characterised their linguistic and geographic distribution, clustered designs into methodological archetypes, modelled archetype membership using multinomial logistic regression, and scored reporting completeness against language-specific items. Here we show that non-English clinical evaluation concentrates in a few high-resource languages. The African and South-East Asian regions (3.15 billion people) contribute 17 of 806 studies (2.1%), with no local-language evaluation for their leading causes of death. Mid-resource languages are more likely than high-resource languages to appear in multi-model examinations rather than clinical utility studies (relative risk ratio 1.79, 95% confidence interval 1.13 to 2.84). Four language-specific items are rarely reported (median 0 of 4) and do not improve over time. The evidence base for clinical language-model safety is misaligned with global deployment and disease burden. We propose LINGUA, a candidate reporting framework that extends existing guidelines to non-English clinical evaluation, and conclude that closing the gap requires external reporting standards and funding priorities, not the passage of time.

  • Decision models can offload structured clinical tasks from large language models

    Dean Ariel, Guy Laban, Stav Endelman, Ran Balicer, Noa Dagan, Seffi Cohen. medRxiv, 2026.

    TL;DR: A low-cost decision model can take over the fully confident share of structured clinical tasks from LLMs, cutting flagship LLM costs by a third or more without losing accuracy.

    Decision models like Jev read text and return probabilities, categories, or ordinal levels, with confidence scores for the latter two, at a fraction of large language models’ (LLMs) cost. We explored how Jev could be integrated into clinical workflows for all these output types: mortality risk estimation, categorical Choosing-Wisely decisions for patient questions, and ordinal colon cancer staging. With standard LLM-style inputs, Jev trailed five flagship LLMs: mortality AUROC was 0.822 versus 0.837–0.846; Choosing-Wisely accuracy, 87.7% versus 98.4–100%; and staging accuracy, 45.4% versus 94.2–100%. Developer-recommended input alterations improved staging accuracy to 75.8% with medical context, 82.1% with task decomposition, and 85.8% with both. None of Jev’s staging errors had full confidence. Using Jev for fully confident staging decisions and LLMs for the rest was non-inferior to each LLM alone and cut flagship LLM costs by 33–44%. Prompted as designed, decision models may efficiently share structured clinical tasks with LLMs.

  • Co-Designing Social Robots for Social-Cognition Training with Autistic Adults

    Yuval Zohar, Mordi Benhamou, Guy Laban. 2026 IEEE 35th International Conference on Robot and Human Interactive Communication (RO-MAN), 2026.

    TL;DR: Five autistic adults co-designed a social robot for social-cognition training. They want a private, configurable rehearsal partner that builds independence, not a companion.

    Social robots have been widely explored as tools for autism intervention, yet this literature has focused predominantly on children and has rarely involved autistic adults as active contributors to design. This creates a mismatch between existing systems and the social-cognitive challenges autistic adults actually face in everyday life, including navigating ambiguous interpersonal contexts, managing conversational timing, and interpreting implied emotional meaning. To address this gap, we conducted an online focus group and co-design session with five autistic adults to explore what a social robot for social-cognition training should do, how it should interact, and under what conditions it would be genuinely useful. The 90-minute session combined open discussion with structured co-design activities on a shared digital whiteboard, and the resulting verbal and visual data were analysed using reflexive thematic analysis. The analysis yielded seven themes that define core design requirements: the robot should function as a scaffold rather than a substitute, prioritise authenticity over comfort, provide personalised and user-controlled feedback, accommodate emotional self-awareness gaps, respect privacy and contextual boundaries, support rehearsal for real-world social situations, and remain configurable in identity, form, and expression. Together, the findings suggest that autistic adults envision the robot not as a companion or live social assistant, but as a private, configurable rehearsal partner designed to support independence over time.

  • Why Did My Robot Just Change Personality? Prompting Guidelines for a Grounded Robot Persona in LLM-Based HRI

    Ashita Ashok, Franziska Babel, Patrick Holthaus, Rucha Khot, Karla Bransky, Fethiye Irmak Dogan, Karsten Berns, Silvia Rossi, Minha Lee, Guy Laban. 2026 IEEE 35th International Conference on Robot and Human Interactive Communication (RO-MAN), 2026.

    TL;DR: LLM-driven robots can show unstable or misleading personas. From a literature review and 27 HRI experts, this paper proposes an eight-part prompt template and guidelines for grounded robot personas.

    Large language models (LLMs) are increasingly used for verbal interaction in social robots, yet prompt design in human-robot interaction (HRI) remains underspecified. As a result, robots may present hallucinated capabilities, unclear behavioural boundaries, and misleading personas. This paper develops a framework for prompt design in LLM-based robots and introduces a structured prompt template comprising eight functional components through which robot behaviour can be specified, bounded, and adapted. The framework is grounded in a review of prior LLM-based HRI work and complemented by survey and discussion data from HRI experts gathered at the Robo-Identity workshop at IEEE RO-MAN $2025 (N=27)$. The qualitative findings highlight limited legibility of robot personality, the need for user adaptation, and strong ethical concerns about safety, deception, and governance. Based on these findings, we present prompting guidelines accompanied by proof-of-concept template as a structured design and reporting aid for HRI research. We argue that prompt design should be treated as a socio-technical problem rather than a minor implementation detail, requiring explicit capability boundaries, transparent behavioural assumptions, and context-sensitive safeguards to support reliable and interpretable HRI.

  • Do Robot-Led Support Conversations Align with Human Therapy Dialogue?

    Sophie Chiang, Guy Laban, Hatice Gunes. 2026 IEEE 35th International Conference on Robot and Human Interactive Communication (RO-MAN), 2026.

    TL;DR: This study investigates how closely robot-led emotional support conversations mirror real human therapy sessions. By highlighting key similarities and communication gaps, the authors provide valuable insights for designing more effective robotic mental health coaches.

    As conversational agents increasingly engage in emotionally supportive dialogue, it is important to understand whether the topics and semantic patterns in robot-led supportive dialogue overlap with those found in human-therapist conversations. This exploratory study investigates whether the topics shared with a robot align with those shared in human-to-human (H2H) therapy sessions, whether subjects use similar semantics when disclosing to these agents, and whether robot responses show semantic proximity to therapist responses within thematically similar disclosures. We analysed two corpora: one of interactions between users and professional therapists (Hugging Face's NLP Mental Health Conversations), and another involving supportive conversations with a social robot (QTrobot; LuxAI) powered by a large language model (LLM; GPT-3.5). Using sentence embeddings and K-means clustering, we assessed cross-agent thematic alignment by applying a distance-based cluster-fitting method that evaluates whether responses from one agent type map to clusters derived from the other, and validating it using Euclidean distances. Results showed that 90.88% of robot conversation disclosures could be mapped to clusters from the human therapy corpus, suggesting shared topical structure. For matched clusters, we compared participant disclosures as well as therapist and robot responses using Sentence Transformer, Word2Vec, and BERT embeddings, revealing model-dependent evidence of semantic overlap in subjects' disclosures and in agent responses to thematically similar disclosures. These findings highlight both the parallels and boundaries of robot-led support conversations and their potential as bounded, complementary tools for emotionally supportive communication.

  • Towards a CONCORD Between Patient Perspective and Ecological Case Formulation: Conversational Large Language Models for Personalised Mental Health

    Nimrod Hertz-Palmor, Amit Oren, George Phillips, Yuxi Wang, Gary Brown, Anna Bevan, Tim Dalgleish, Guy Laban. Research Square, 2026.

    TL;DR: Three weeks of daily conversations with an AI built personalised causal maps of PTSD for 11 people that matched their own accounts and leading clinical theory.

    Understanding the mechanisms that maintain mental disorders at the individual level is essential for personalised psychological treatment, yet existing approaches rely on burdensome questionnaires or statistical models that infer causal relationships indirectly. Here we present CONCORD, an AI–supported conversational framework that continuously constructs individualised causal case formulations from natural patient conversations. Eleven people with posttraumatic stress disorder engaged in daily conversations over three weeks while also completing ecological momentary assessments and causal surveys for validation, generating 1,629 interactions across methods. CONCORD recovered individualised causal structures that closely matched patient-reported causal formulations while substantially exceeding the complexity captured by conventional correlational network models. The resulting networks closely conformed to a leading cognitive theory of posttraumatic stress disorder, accurately identifying trauma-related triggers as primary causal drivers and maladaptive coping strategies as downstream consequences. Because each inferred relationship is traceable to the underlying conversation, the framework remains transparent and clinically interpretable despite relying on AI. These findings demonstrate that natural conversation can provide a practical route to continuous, theory-guided case formulation, offering a scalable framework for ecological patient monitoring and personalised precision mental health.

  • Underwriting the Agent Economy: The Blueprint for an AI Insurance Stack

    Cristian Trout, Sanmi Koyejo, Sasha Romanosky, Giorgio Ripamonti, Lynn Thompson, Desiree Spain, Alex Taylor, Kevin Casey, Stephen Casper, Matthew Botvinick, Sean McGregor, Miles Brundage, A. Feder Cooper, Patricia Paskov, Adrien Ecoffet, Ben Bucknall, Kevin Wei, Markus Anderljung, Lukasz Szpruch, Bri Treece, Tom Zick, Gabriel Weil, Ugur Ozer, Kevin Kalinich, Jesus Gonzalez, Vitaly Baranov, Moran Koren, Guy Laban, Gil Arazi, Henri Winand, Derek Blum, Toby Clowes, Adam Kleinman, Anita Srinivasan, Tom Fehring, Rune Kvist, Rajiv Dattani. arXiv, 2026.

    TL;DR: A blueprint for insuring AI agents: an eight-part insurance stack, from incident data to claims handling, to cover AI agent risk at scale by 2030.

    From maritime trade to commercial nuclear power, insurance has been the enabler of major economic and technological developments by pricing risk, limiting downside, and spreading best practices. The emerging AI agent economy, projected to handle trillions of dollars in transactions by 2030, looks to be the next such development. Yet insurers' exposure to AI agent risk currently sits largely unpriced across existing insurance lines; between this silent coverage and growing exclusions, coverage is not fit for purpose. Furthermore, insurability is trending the wrong way: AI agent capabilities appear to be outpacing reliability, leading to rising incident severity; concentration among a few foundation model providers threatens correlated losses; and traditional actuarial modeling will struggle to keep pace with a technology evolving as rapidly as frontier AI. This report argues that affirmative AI coverage with limits in the billions is achievable by 2030, but only with industry-wide coordination. Drawing on successful historical precedents such as Underwriters Laboratories, the Closed Claims Project, and others, we lay out an eight-component AI insurance stack spanning incident data collection, catastrophe modeling, standards, contract design, risk selection, pricing, monitoring, and claims management. Building out this infrastructure is what will enable insurers to cover and manage AI agent risk sustainably and at scale. Finally, we discuss coverage for catastrophic risk from frontier AI ("AI CAT"), including CBRN, critical infrastructure collapse, and loss of control scenarios. Addressing these tail risks will require purpose-built instruments, potentially including a frontier model developer mutual, catastrophe bonds, bespoke liability regimes, and government backstops.

  • SPASM: Stable Persona-driven Agent Simulation for Multi-turn Dialogue Generation

    Han Luo, Guy Laban. Findings of the Association for Computational Linguistics: ACL 2026, 2026.

    TL;DR: This paper introduces SPASM, a framework that generates stable, persona-consistent multi-turn dialogues for LLMs by using a novel "Egocentric Context Projection" method to prevent identity-related failures like persona drift and echoing.

    Large language models are increasingly deployed in multi-turn settings such as tutoring, support, and counseling, where reliability depends on preserving consistent roles, personas, and goals across long horizons. This requirement becomes critical when LLMs are used to generate synthetic dialogues for training and evaluation, since LLM--LLM conversations can accumulate identity-related failures such as persona drift, role confusion, and "echoing", where one agent gradually mirrors its partner. We introduce SPASM (Stable Persona-driven Agent Simulation for Multi-turn dialogue generation), a modular, stability-first framework that decomposes simulation into (i) persona creation via schema sampling, plausibility validation, and natural-language persona crafting, (ii) Client--Responder dialogue generation, and (iii) termination detection for coherent stopping. To improve long-horizon stability without changing model weights, we propose Egocentric Context Projection (ECP): dialogue history is stored in a perspective-agnostic representation and deterministically projected into each agent's egocentric view before generation. Across three LLM backbones (GPT-4o-mini, DeepSeek-V3.2, Qwen-Plus) and nine Client--Responder pairings, we construct a dataset of 4,500 personas and 45,000 conversations (500 personas X 10 conversations per pairing). Ablations show ECP substantially reduces persona drift and, under human validation, eliminates echoing; embedding analyses recover persona structure and reveal strong responder-driven interaction geometry. Our code is available at this https URL.

  • DialogGuard: Multi-Agent Psychosocial Safety Evaluation Interface of Sensitive LLM Responses

    Han Luo, Guy Laban. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 3: System Demonstrations), 2026.

    TL;DR: This paper introduces DialogGuard, a system that uses multiple AI agents to evaluate the psychosocial safety of LLM responses in sensitive areas like mental health, identifying risks like privacy violations or psychological harm more accurately than current methods. It helps ensure AI models are safer for vulnerable users by providing detailed risk scores and explanations.

    Large language models (LLMs) now mediate many web-based mental- health, crisis, and other emotionally sensitive services, yet their psychosocial safety in these settings remains poorly understood and weakly evaluated. We present DialogGuard, a multi-agent frame- work for assessing psychosocial risks in LLM-generated responses along five high-severity dimensions: privacy violations, discrimi- natory behaviour, mental manipulation, psychological harm, and insulting behaviour. DialogGuard can be applied to diverse gen- erative models through four LLM-as-a-judge pipelines, including single-agent scoring, dual-agent correction, multi-agent debate, and stochastic majority voting, grounded in a shared three-level rubric usable by both human annotators and LLM judges. Using PKU-SafeRLHF with human safety annotations, we show that multi- agent mechanisms detect psychosocial risks more accurately than non-LLM baselines and single-agent judging; dual-agent correction and majority voting provide the best trade-off between accuracy, alignment with human ratings, and robustness, while debate attains higher recall but over-flags borderline cases. We release Dialog- Guard as open-source software with a web interface that provides per-dimension risk scores and explainable natural-language ratio- nales. A formative study with 12 practitioners illustrates how it supports prompt design, auditing, and supervision of web-facing applications for vulnerable users.

  • Designing Artificial Identity: The Identity Design Framework and Research Agenda

    Karla Bransky, Penny Sweetser, Patrick Holthaus, Guy Laban, Katie Winkle, Neziha Akalin, Ashita Ashok, Jihye Lee, Rucha Khot, Alexandra Bejarano, Jorrit Thijn, Roger K. Moore, Minsu Jang, Joel E Fischer, Minha Lee. Proceedings of the 2026 Designing Interactive Systems Conference, 2026.

    TL;DR: This paper introduces the "Identity Design (ID) Framework," offering a structured approach with twelve principles to help designers create coherent and ethical artificial identities for AI and robots. It aims to prevent issues like harmful stereotypes and deception by guiding how artificial agents present themselves as individuals, members of groups, and societal entities.

    The identity design of artificial agents carries growing ethical, psychological, and cultural weight, as ubiquitous language models and diverse robotic forms are blended into everyday use. However, structured approaches to designing coherent and interpretable artificial identities remain limited. To address urgent challenges in artificial identity design, including harmful stereotypes and deceptive practices, we introduce the Identity Design (ID) Framework and an accompanying research agenda. Drawing on emerging work on artificial identity in human-robot interaction and taking an interdisciplinary perspective, we propose twelve design principles across three levels: individual (recognisability, behavioural consistency, identity continuity, memory, persistent goals), group (membership signalling, social alignment, role clarity), and societal (benevolence, artificiality, social justice, transparency). The research agenda outlines open questions around the operationalisation and measurement of identity, social dynamics, and ethical considerations for identity design. Together, they lay the groundwork for future research and responsible practice in robotic, virtual, and multi-embodied agents.

  • Consumer chatbots gave similarly empathic answers whether safe or unsafe: a physician-rated evaluation in six languages

    Dean Ariel, Lyel Romina Grumberg, Sopak Supakul, Sirawit Wannasri, Ilan Y. Mitchnik, Anna Lev, Weerawat Ariyamethanon, Muhammad Agbarieh, Shafiq Miari, Guy Laban, Boaz Hasid. medRxiv, 2026.

    TL;DR: Physicians rated 504 chatbot answers in six languages. How empathic an answer sounded said nothing about whether it was safe, and clinical substance dropped in Hebrew.

    Background Patients cannot check the clinical content of chatbot health advice, so they judge it by what they can perceive. We examined whether physician-rated empathy tracked clinical quality, and whether that relationship held when the question was asked in another language. Methods Four consumer chatbots answered forum-derived, clinician-adapted patient scenarios in six languages (English, Hebrew, French, Russian, Arabic, Thai), yielding 504 responses. Two language-matched physicians per language, blinded to chatbot identity, rated accuracy, safety, referral, cultural appropriateness, and empathy, and completed an item-level checklist, giving 1,008 ratings across 21 scenarios. Associations were estimated within physicians. The same two physicians rated English and Hebrew, so that contrast was also within-physician. Results Empathy did not separate safe from unsafe responses (AUC 0.49, 95% CI 0.39 to 0.62), and within-physician slopes on safety and substance were near zero (−0.006 and −0.004). Each dimension correlated with its own checklist (r = 0.81 and 0.70) and not with the other (0.01 and −0.01). The substance-minus-empathy gap narrowed from 0.92 to 0.44 and from 1.09 to 0.52 in the two English– Hebrew physicians, driven by lower substance. Unsafe ratings concentrated on the same three scenarios across products (p<0.001), and ten responses were accurate yet unsafe. Conclusions Empathy, the one cue a patient can judge, carried no information about whether the advice was safe, and clinical substance fell in Hebrew within both physicians who rated it. Evaluation should score clinical content independently of empathy, in each deployment language, and anchor on high-risk scenarios rather than any single product.

  • Emotional Expression in Low-Degrees-of-Freedom Robots: Assessing Perception with Reachy Mini

    Amit Rogel, Elmira Yadollahi, Guy Laban. arXiv, 2026.

    TL;DR: 100 people watched a minimal robot express ten emotions. Exact labels were hard to read, but the broad feeling came through and shaped how warm and sociable the robot seemed.

    Emotion expression is central to human--robot interaction, yet little is known about how people interpret affect on robots with sparse, non-anthropomorphic expressive capabilities. This study examined how people perceive emotional expressions displayed by Reachy Mini (Pollen Robotics and Hugging Face), a low-degree-of-freedom (low-DoF) robot with a constrained and distinctly non-human expressive repertoire. In an online within-subjects study, 100 participants viewed 10 short video clips of Reachy Mini expressing different emotions and, for each clip, identified the perceived emotion, rated its valence and arousal, and evaluated the robot on social-perception traits. Exact emotion recognition was modest overall and varied considerably across expressions, with anger, sadness, and interest recognized more reliably than emotions such as love, pleasure, shame, and disgust. However, participants were generally more successful at recovering broader affective meaning than exact emotion labels, particularly along valence and arousal dimensions. Emotional expressions also shaped social evaluation, as positive expressions were perceived as warmer and more sociable than negative ones, and animacy varied less across conditions. These findings suggest that even constrained robotic expressions can communicate affective meaning and influence social impressions, positioning Reachy Mini as a useful benchmark for studying affective communication in low-DoF robots.

  • Social Robotics for Disabled Students: An Empirical Investigation of Embodiment, Roles, and Interaction

    Alva Markelius, Fethiye Irmak Dougan, Julie Bailey, Guy Laban, Jenny L. Gibson, Hatice Gunes. Proceedings of the 21st ACM/IEEE International Conference on Human-Robot Interaction, 2026.

    TL;DR: This study found that disabled students perceived physical robots as more understanding than voice-only agents for support, highlighting the importance of robot embodiment and how different robot roles and interaction styles impact accessibility in higher education.

    Institutional and social barriers in higher education often prevent students with disabilities from effectively accessing support, including lengthy procedures, insufficient information, and high social-emotional demands. This study empirically explores how disabled students perceive robot-based support, comparing two interaction roles, one information based (signposting) and one disclosure based (sounding board), and two embodiment types (physical robot/disembodied voice agent). Participants assessed these systems across five dimensions: perceived understanding, social energy demands, information access/clarity, task difficulty, and data privacy concerns. The main findings of the study reveal that the physical robot was perceived as more understanding than the voice-only agent, with embodiment significantly shaping perceptions of sociability, animacy, and privacy. We also analyse differences between disability types. These results provide critical insights into the potential of social robots to mitigate accessibility barriers in higher education, while highlighting ethical, social and technical challenges.

  • A Robot-Led Intervention for Emotion Regulation: from Expression to Reappraisal

    Guy Laban, Julie Wang, Hatice Gunes. IEEE Transactions on Affective Computing, 2026.

    TL;DR: This paper develops and tests a robot that helps people learn to better manage their emotions using techniques like expressing feelings and changing their thoughts about situations.

    Emotion regulation (ER) is a crucial skill for managing emotions in everyday life, yet finding a constructive and accessible method to practice ER remains challenging due to cognitive demands. In this study, we explore how regular interactions with a social robot, conducted in a structured yet familiar environment within university halls and departments, can provide effective support for practising ER via cognitive reappraisal. Twenty-one students participated in a five-session study at a university hall or department in the U.K., where the robot (QTrobot, LuxAI), powered by a large language model (GPT 3.5) facilitated structured conversations, encouraging the students to reinterpret emotionally charged situations that they shared with the robot. Quantitative and qualitative results indicate significant improvements in ER, with participants reporting better understanding and control of their emotions, significant changes in constructive ER tendencies and positive effects on mood and sentiment after each session. The findings also demonstrate that repeated interactions with the robot encouraged greater emotional expression, including longer speech disclosures, increased use of affective language, and heightened facial arousal. Notably, expression peaked during key reappraisal moments, particularly when participants were prompted to reinterpret negative experiences. The qualitative feedback further highlighted how the robot fostered introspection and provided a supportive space for discussing emotions, enabling participants to confront long-avoided emotional challenges. These findings demonstrate the potential of robots to effectively assist in ER in familiar environments, offering both emotional support and cognitive guidance

2025

  • A Multimodal Neural Network for Recognizing Subjective Self-Disclosure Towards Social Robots

    Henry Powell, Guy Laban, Emily S. Cross. 2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2025.

    TL;DR: This paper introduces a new AI that can identify when people are sharing personal information with social robots by analyzing their speech and non-verbal cues. This helps robots better understand and respond appropriately to human vulnerability.

    Subjective self-disclosure is an important feature of human social interaction. While much has been done in the social and behavioural literature to characterise the features and consequences of subjective self-disclosure, little work has been done thus far to develop computational systems that are able to accurately model it. Even less work has been done that attempts to model specifically how human interactants self-disclose with robotic partners. It is becoming more pressing as we require social robots to work in conjunction with and establish relationships with humans in various social settings. In this paper, our aim is to develop a custom multimodal attention network based on models from the emotion recognition literature, training this model on a large self-collected self-disclosure video corpus, and constructing a new loss function, the scale preserving cross entropy loss, that improves upon both classification and regression versions of this problem. Our results show that the best performing model, trained with our novel loss function, achieves an F1 score of 0.83, an improvement of 0.48 from the best baseline model. This result makes significant headway in the aim of allowing social robots to pick up on an interaction partner’s self-disclosures, an ability that will be essential in social robots with social cognition.

  • DEBP-PVA: Designing and Evaluating Behavioral Paradigms With Proactive Virtual Agents

    Alafate Abulimiti, Paola R Peña, Fatemeh Alizadeh, Shashank Ahire, Heloisa Candello, Smit Desai, Justin Edwards, Yuan He, Darragh Higgins, Alberto Jovane, Matthias Kraus, Guy Laban, Rachel McDonnell, Jairo Pérez-Osorio, Tanja Schneeberger, Jaisie Sin, Tobias Thejll-Madsen, Nima Zargham, Benjamin R Cowan. Adjunct Proceedings of the 25th ACM International Conference on Intelligent Virtual Agents, 2025.

    TL;DR: A workshop on designing and evaluating behavioural paradigms that elicit and measure proactive behaviour in intelligent virtual agents.

    This workshop examines the design and evaluation of behavioral paradigms that effectively elicit and measure proactive behavior in Intelligent Virtual Agents (IVA). Current paradigms focus mainly on agent behavior. However, it remains unclear how these paradigms effectively help to elicit and measure proactive agent interactions and produce ecologically valid data, particularly for paradigms that promote the use of different modalities during interaction beyond text or speech. We invite researchers to discuss methodologies for developing tasks that not only structure agent behaviors but also provide meaningful insights into the social and cognitive processes involved in proactive interactions.

  • What People Share With a Robot When Feeling Lonely and Stressed and How It Helps Over Time

    Guy Laban, Sophie Chiang, Hatice Gunes. 2025 34th IEEE International Conference on Robot and Human Interactive Communication (RO-MAN), 2025.

    TL;DR: This study found that young adults who repeatedly interacted with a robot powered by AI reported feeling less lonely and stressed. The types of things they talked about with the robot changed based on their initial stress levels, with more distressed individuals discussing social connections and less distressed individuals focusing on personal goals.

    Loneliness and stress are prevalent among young adults and are linked to significant psychological and health-related consequences. Social robots may offer a promising avenue for emotional support, especially when considering the ongoing advancements in conversational AI. This study investigates how repeated interactions with a social robot influence feelings of loneliness and perceived stress, and how such feelings are reflected in the themes of user disclosures towards the robot. Participants engaged in a five-session robot-led intervention, where a LLM-powered QTrobot facilitated structured conversations designed to support cognitive reappraisal. Results from linear mixed-effects models show significant reductions in both loneliness and perceived stress over time. Additionally, semantic clustering of 560 user disclosures towards the robot revealed six distinct conversational themes. Results from Kruskal-Wallis H-test demonstrate that participants reporting higher loneliness and stress, more frequently engaged in socially focused disclosures, such as friendship and connection, whereas lower distress was associated with introspective and goal-oriented themes (e.g., academic ambitions). By exploring both how the intervention affects well-being, as well as how well-being shapes the content of robot-directed conversations, we aim to capture the dynamic nature of emotional support in human–robot interaction.

  • Robot-Led Vision Language Model Wellbeing Assessment of Children

    Nida Itrat Abbasi, Fethiye Irmak Dogan, Guy Laban, Joanna Anderson, Tamsin Ford, Peter B. Jones, Hatice Gunes. 2025 34th IEEE International Conference on Robot and Human Interactive Communication (RO-MAN), 2025.

    TL;DR: A robot used AI to assess children's mental well-being by analyzing their stories about pictures, showing promise for identifying healthy children but struggling with concerns and exhibiting gender bias.

    This study presents a novel robot-led approach to assessing children’s mental wellbeing using a Vision Language Model (VLM). Inspired by the Child Apperception Test (CAT), the social robot NAO presented children with pictorial stimuli to elicit their verbal narratives of the images, which were then evaluated by a VLM in accordance with CAT assessment guidelines. The VLM’s assessments were systematically compared to those provided by a trained psychologist. The results reveal that while the VLM demonstrates moderate reliability in identifying cases with no wellbeing concerns, its ability to accurately classify assessments with wellbeing concerns remains limited. Moreover, although the model’s performance was generally consistent when prompted with varying demographic factors such as age and gender, a significantly higher false positive rate was observed for girls, indicating potential sensitivity to gender attribute. These findings highlight both the promise and the challenges of integrating VLMs into robot-led assessments of children’s wellbeing.

  • Self-Disclosure Themes and Semantics Across Human, Robotic, and Disembodied Conversational Partners

    Sophie Chiang, Guy Laban, Emily S. Cross, Hatice Gunes. 2025 34th IEEE International Conference on Robot and Human Interactive Communication (RO-MAN), 2025.

    TL;DR: People disclose the same themes, in semantically similar ways, whether they talk to a human, a humanoid robot or a disembodied agent.

    As social robots and other artificial agents become more conversationally capable, it is important to understand whether the content and meaning of self-disclosure towards these agents changes depending on the agent’s embodiment. In this study, we analysed conversational data from three controlled experiments in which participants self-disclosed to a human, a humanoid social robot, and a disembodied conversational agent. Using sentence embeddings and clustering, we identified themes in participants’ disclosures, which were then labelled and explained by a large language model. We subsequently assessed whether these themes and the underlying semantic structure of the disclosures varied by agent embodiment. Our findings reveal strong consistency: thematic distributions did not significantly differ across embodiments, and semantic similarity analyses showed that disclosures were expressed in highly comparable ways. These results suggest that while embodiment may influence human behaviour in human–robot and human–agent interactions, people tend to maintain a consistent thematic focus and semantic structure in their disclosures, whether speaking to humans or artificial interlocutors.

  • How Large Language Models Classify and Semantically Explain Facial Expressions from Valence-Arousal Values

    Vaibhav Mehra, Guy Laban, Hatice Gunes. CUI 2025 - Proceedings of the 2025 ACM Conference on Conversational User Interfaces, 2025.

    TL;DR: Large Language Models (LLMs) can't reliably classify emotions from numerical representations of facial expressions, but they are surprisingly good at describing a face's emotional meaning in words, similar to how humans would.

    Large Language Models (LLMs) primarily operate through text-based inputs and outputs, yet emotion is communicated through both verbal and non-verbal cues, including facial expressions. While Vision-Language Models analyse facial expressions from images, they are resource-intensive and may depend more on linguistic priors than visual understanding. To address this, this study investigates whether LLMs can infer affective meaning from dimensions of facial expressions—Valence-Arousal (VA) values, structured numerical representations. VA values were extracted using Facechannel from images of facial expressions (from IIMI and EMOTIC datasets) and provided to three LLMs in two tasks: (1) classifying facial expressions into basic and complex emotions and (2) generating semantic descriptions of facial expressions. The results indicate that LLMs struggle to classify VA values into discrete emotion categories, particularly for emotions beyond basic polarities. However, LLMs produced semantic descriptions that align closely with human-generated interpretations, demonstrating a stronger capacity for free-text affective inference of facial expressions.

  • Bias and Fairness in Conversational User Interfaces

    Guy Laban, Julian Hough, Minha Lee, Alva Markelius, Mary Ellen Foster, Jane Stuart-Smith, Muneeb Imtiaz Ahmad. Proceedings of the 7th ACM Conference on Conversational User Interfaces, 2025.

    TL;DR: A CUI 2025 workshop that treats bias in chatbots, virtual agents and social robots as a relational and design problem, not only a technical flaw.

    Conversational User Interfaces (CUIs), including chatbots, virtual agents and social robots, are increasingly shaping how we communicate, seek support and access services. Yet, as these systems grow more sophisticated, concerns about bias and fairness in their design and deployment have become increasingly urgent. We propose a multidimensional approach to bias and fairness in CUIs that spans four interconnected themes: conceptual grounding, verbal communication, multimodal expression and interactional dynamics. Rather than framing bias merely as a technical flaw, we argue that it should be understood as a relational, interactional and design-based phenomenon. Accordingly, in this workshop, we aim to foster critical discussion around how CUIs encode social norms, perpetuate or mitigate exclusion, and shape perceptions of fairness through their language, embodiment and behaviour. By bringing together researchers, designers and policymakers, the workshop will explore pathways towards more equitable and transparent CUIs. The goal is to promote a relational understanding of fairness, one that centres user experience and social context, to guide future work in conversational AI.

  • Knowledge cues to human origins facilitate self-disclosure during interactions with chatbots

    Gabriella Warren-Smith, Guy Laban, Emily-Marie Pacheco, Emily S. Cross. Computers in Human Behavior: Artificial Humans, 2025.

    TL;DR: People tend to disclose more personal information to chatbots they believe are human, but paradoxically, they find chatbots they *know* are AI to be friendlier and more sentimental. This suggests that honesty about a chatbot's AI nature might build a different kind of trust.

    Chatbots are emerging as a self-management tool for supporting mental health, appearing across commercial and healthcare settings. Whilst chatbots are valued for their perceived lack of judgement, they lack the emotional intelligence and empathy to build trust and rapport with users. A resulting debate questions whether chatbots facilitate or hinder self-disclosure. This study presents a within-subjects experimental design investigating the parameters of self-disclosure in social interactions with chatbots in an open domain. Participants engaged in two short social interactions with two chatbots: one with the knowledge they were conversing with a chatbot and one with the false belief they were conversing with a human. A significant difference was found across both treatments, with participants disclosing more to the chatbot that was introduced as a human, as well as perceiving themselves to do so, perceiving this chatbot as more comforting, and to be demonstrating higher rates of agency and experience compared to the chatbot that was introduced as a chatbot. However, significant findings also indicated participants' disclosures to the chatbot that was introduced as a chatbot were more sentimental, and they found it to be friendlier compared to the chatbot that was introduced as a human. These results indicate that whilst cues to a chatbot’s human origins enhance self-disclosure and perceptions of mind, when the artificial agent is perceived against one’s social expectations, it may be viewed negatively on social factors that require higher cognitive processing.

  • Critical Insights about Robots for Mental Wellbeing

    Guy Laban, Micol Spitale, Minja Axelsson, Nida Itrat Abbasi, Hatice Gunes. arXiv, 2025.

    TL;DR: Six lessons from studies and deployments of robots for mental wellbeing, including that robots need not be companions to help and should support, not replace, therapists.

    Social robots are increasingly being explored as tools to support emotional wellbeing, particularly in non-clinical settings. Drawing on a range of empirical studies and practical deployments, this paper outlines six key insights that highlight both the opportunities and challenges in using robots to promote mental wellbeing. These include (1) the lack of a single, objective measure of wellbeing, (2) the fact that robots don't need to act as companions to be effective, (3) the growing potential of virtual interactions, (4) the importance of involving clinicians in the design process, (5) the difference between one-off and long-term interactions, and (6) the idea that adaptation and personalization are not always necessary for positive outcomes. Rather than positioning robots as replacements for human therapists, we argue that they are best understood as supportive tools that must be designed with care, grounded in evidence, and shaped by ethical and psychological considerations. Our aim is to inform future research and guide responsible, effective use of robots in mental health and wellbeing contexts.

  • Coping with Emotional Distress via Self-Disclosure to Robots: An Intervention with Caregivers

    Guy Laban, Val Morrison, Arvid Kappas, Emily S. Cross. International Journal of Social Robotics, 2025.

    TL;DR: Talking to a robot can help caregivers feel less stressed and lonely over time, as they increasingly disclose their feelings and find the robot comforting. This suggests robots could be a new source of emotional support for those in demanding caregiving roles.

    People often engage in self-disclosure and social sharing when trying to cope with emotional distress. This study introduces a novel long-term intervention designed to help informal caregivers cope with emotional distress by self-disclosing towards a social robot. Research indicates that informal caregivers frequently face challenges in handling the emotional and practical demands of caregiving, often experiencing a lack of social support and limited social interaction. Accordingly, we explored the extent of informal caregivers’ self-disclosure behaviour towards a social robot (Pepper, SoftBank Robotics) over time, and how their perceptions of the robot evolved. Additionally, we examined how this intervention affected caregivers’ moods, perceptions of the robot as comforting, feelings of loneliness, stress levels, as well as its impact on their emotion regulation. We replicated a previous long-term experiment [1] with a dedicated sample of informal caregivers who interacted with Pepper 10 times over five weeks, discussing everyday topics. Our results show that caregivers increasingly self-disclosed to the robot over time, perceiving it as more social and competent. Participants’ moods improved following interactions, and they viewed the robot as increasingly comforting. They also reported feeling progressively less lonely and stressed. Thus, our findings with informal caregivers replicated those of [1]. Moreover, after the intervention, caregivers reported greater acceptance of their caregiving roles, reappraising it more positively, and reduced feelings of blame towards others. These results highlight the potential of social robots to provide emotional support for individuals coping with emotional distress.

  • A Longitudinal Study of Child Wellbeing Assessment via Online Interactions with a Social Robot

    Nida Itrat Abbasi, Guy Laban, Tamsin Ford, Peter B. Jones, Hatice Gunes. ACM Transactions on Human-Robot Interaction, 2025.

    TL;DR: This study found that children's mental well-being can be assessed over time through online interactions with a robot, and children generally maintained positive perceptions of the robot. The research suggests that robots could be a valuable tool for supporting children's mental health remotely.

    Socially Assistive Robots are studied in different child-robot interaction settings. However, logistical constraints limit accessibility, particularly affecting timely support for mental wellbeing. In this work, we have investigated whether online interactions with a robot can be used for the assessment of mental wellbeing in children. The children (N = 40, 20 girls and 20 boys; 8-13 years) interacted with the Nao robot (30-45 mins) over three sessions, at least a week apart. Audio-visual recordings were collected throughout the sessions that concluded with the children answering user perception questionnaires pertaining to their anxiety toward the robot, and the robot's abilities. We divided the participants into three wellbeing clusters (low, med, and high tertiles) using their responses to the Short Moods and Feelings Questionnaire (SMFQ) and further analyzed how their wellbeing and their perceptions of the robot changed over the wellbeing tertiles, across sessions and across participants' gender. Our primary findings suggest that (I) online-mediated interactions with robots can be effective in assessing children's mental wellbeing over time, and (II) children's overall perception of the robot either improved or remained consistent across time. Supplementary exploratory analyses have also revealed that the gender of the children affected their wellbeing assessments with interactions effectively distinguishing between varying levels of wellbeing for both boys and girls for the first session and only for boys during the second session. The analyses have also revealed that girls have a higher opinion of the robot as a confidante as compared with boys. Findings from this work affirm the potential of using online-mediated interactions with robots for the assessment of the mental wellbeing of children.

  • Past, Present, and Future: A Survey of The Evolution of Affective Robotics For Well-being

    Micol Spitale, Minja Axelsson, Sooyeon Jeong, Paige Tuttösí, Caitlin A. Stamatis, Guy Laban, Angelica Lim, Hatice Gunes. IEEE Transactions on Affective Computing, 2025.

    TL;DR: This paper surveys how robots have evolved to understand and respond to human emotions, and how they can be used in the future to improve our well-being.

    Recent research in affective robots has recognized their potential in supporting human well-being. Due to rapidly developing affective and artificial intelligence technologies, this field of research has undergone explosive expansion and advancement in recent years. In order to develop a deeper understanding of recent advancements, we present a systematic review of the past 10 years of research in affective robotics for wellbeing. In this review, we identify the domains of well-being that have been studied, the methods used to investigate affective robots for well-being, and how these have evolved over time. We also examine the evolution of the multifaceted research topic from three lenses: technical, design, and ethical. Finally, we discuss future opportunities for research based on the gaps we have identified in our review – proposing pathways to take affective robotics from the past and present to the future. The results of our review are of interest to human-robot interaction and affective computing researchers, as well as clinicians and well-being professionals who may wish to examine and incorporate affective robotics in their practices.

2024

  • LEXI: Large Language Models Experimentation Interface

    Guy Laban, Tomer Laban, Hatice Gunes. Proceedings of the 12th International Conference on Human-Agent Interaction, 2024.

    TL;DR: This paper introduces LEXI, an open-source tool that helps researchers easily build and run experiments with AI agents powered by large language models, making it easier to study how people interact with AI. It was successfully used to show that empathetic AI agents are perceived as more social and receive more positive responses.

    The recent developments in Large Language Models (LLMs) mark a significant moment in the research and development of social interactions with artificial agents. These agents are widely deployed in a variety of settings, with potential impact on users. However, the study of social interactions with agents powered by LLMs is still emerging, limited by access to the technology and to data, the absence of standardised interfaces, and challenges to establishing controlled experimental setups using the currently available platforms. To address these gaps, we developed LEXI, LLMs Experimentation Interface, an open-source tool for deploying artificial agents powered by LLMs in social interaction behavioural experiments. Using a graphical interface, LEXI allows researchers to build agents and deploy them in experimental setups along with forms for collecting self-reported data while collecting interaction logs. The outcomes of usability testing indicate LEXI’s broad utility, high usability, and minimal mental workload requirement, with benefits observed across disciplines. A proof-of-concept study exploring the tool’s efficacy in evaluating social human–agent interactions was conducted, resulting in high-quality data. A comparison of empathetic versus neutral agents indicated that people perceive empathetic agents as more social, and write longer and more positive messages towards them.

  • Sharing our Emotions with Robots: Why do we do it and how does it make us feel?

    Guy Laban, Emily S. Cross. IEEE Transactions on Affective Computing, 2024.

    TL;DR: People tend to share their emotions with robots, and this theoretical paper explores why they do it and how it makes them feel, revealing insights into human-robot emotional interaction.

    Self-disclosure and the social sharing of emotions facilitate social relationships and can positively affect people's well-being. Nevertheless, individuals might refrain from engaging in these interpersonal communication behaviours with other people, due to socio-emotional barriers, such as shame and stigma. Social robots, arguably free from these human-centric judgements, could encourage openness and overcome these barriers. Accordingly, this paper reviews the role of self-disclosure and social sharing of emotion in human–robot interactions (HRIs), particularly the implications for emotional well-being and the dynamics of social relationship building between humans and robots. We explore the transition of self-disclosure dynamics from traditional human-to-human interactions to HRI, revealing the potential of social robots to bridge socio-emotional barriers and provide unique forms of emotional support. This review not only highlights the therapeutic potential of robots but also raises critical ethical considerations and potential drawbacks of these interactions, emphasising the importance of a balanced approach to integrating robots into emotional support roles. As a whole, this review introduces a complex but promising frontier at the intersection of technology and emotional well-being, and we advocate for careful consideration of ethical standards and the intrinsic human need for connection as the development and application of social robots continues to advance.

  • Robotising Psychometrics: Validating Wellbeing Assessment Tools in Child-Robot Interactions

    Nida Itrat Abbasi, Guy Laban, Tamsin Ford, Peter B. Jones, Hatice Gunes. IEEE International Workshop on Robot and Human Communication, RO-MAN, 2024.

    TL;DR: This study found that robots can reliably and validly administer mental health questionnaires to children, and sometimes even better than self-reporting, though some questions might need tweaking for robot-led interactions.

    The interdisciplinary nature of Child-Robot Interaction (CRI) fosters incorporating measures and methodologies from many established domains. However, when employing CRI approaches to sensitive avenues of health and wellbeing, caution is critical in adapting metrics to retain their safety standards and ensure accurate utilisation. We conducted a secondary analysis to previous empirical work, investigating the reliability and construct validity of established psychological questionnaires such as the Short Moods and Feelings Questionnaire (SMFQ) and three subscales (generalised anxiety, panic and low mood) of the Revised Child Anxiety and Depression Scale (RCADS) within a CRI setting for the assessment of mental wellbeing. Through confirmatory principal component analysis, we have observed that these measures are reliable and valid in the context of CRI. Furthermore, our analysis revealed that scales communicated by a robot demonstrated a better fit than when self-reported, underscoring the efficiency and effectiveness of robot-mediated psychological assessments in these settings. Nevertheless, we have also observed variations in item contributions to the main factor, suggesting potential areas of examination and revision (e.g., relating to physiological changes, inactivity and cognitive demands) when used in CRI. Our findings highlight the importance of verifying the reliability and validity of standardised metrics and assessment tools when employed in CRI settings, thus, aiming to avoid any misinterpretations and misrepresentations.

  • Studying and Eliciting Self-Disclosure: Interdisciplinary Review of Research Methodologies and Behavioural Paradigms

    Guy Laban. OSF, 2024.

    TL;DR: This paper reviews all the different ways researchers study and measure how people share personal information, highlighting that understanding this complex behaviour requires using many different research methods.

    This paper presents a comprehensive review of methodologies for eliciting, monitoring, and measuring self-disclosure, a critical component of human interaction. It examines both experimental and naturalistic techniques, ranging from interview-based approaches to digital communication analysis, and discusses their advantages and limitations in various contexts. Additionally, it explores various self-disclosure measurement tools, including self-reports, observer ratings, and automated content analysis. The study highlights the complexity of self-disclosure, shaped by individual, relational, and situational factors, and underscores the importance of employing diverse, interdisciplinary methods to accurately capture its dynamics and effects on relationship establishment and maintenance, well-being, and social connections.

  • Building Long-Term Human–Robot Relationships: Examining Disclosure, Perception and Well-Being Across Time

    Guy Laban, Arvid Kappas, Val Morrison, Emily S. Cross. International Journal of Social Robotics, 2024.

    TL;DR: Interacting with a robot over several weeks can lead people to share more, view the robot more positively, and even experience improved mood and reduced loneliness. However, discussing emotionally charged topics with the robot can increase feelings of loneliness and stress.

    While interactions with social robots are novel and exciting for many people, one concern is the extent to which people’s behavioural and emotional engagement might be sustained across time, since during initial interactions with a robot, its novelty is especially salient. This challenge is particularly noteworthy when considering interactions designed to support people’s well-being, with limited evidence (or empirical exploration) of social robots’ capacity to support people’s emotional health over time. Accordingly, our aim here was to examine how long-term repeated interactions with a social robot affect people’s self-disclosure behaviour toward the robot, their perceptions of the robot, and how such sustained interactions influence factors related to well-being. We conducted a mediated long-term online experiment with participants conversing with the social robot Pepper 10 times over 5 weeks. We found that people self-disclose increasingly more to a social robot over time, and report the robot to be more social and competent over time. Participants’ moods also improved after talking to the robot, and across sessions, they found the robot’s responses increasingly comforting as well as reported feeling less lonely. Finally, our results emphasize that when the discussion frame was supposedly more emotional (in this case, framing questions in the context of the COVID-19 pandemic), participants reported feeling lonelier and more stressed. These results set the stage for situating social robots as conversational partners and provide crucial evidence for their potential inclusion in interventions supporting people’s emotional health through encouraging self-disclosure.

2023

  • Opening Up to Social Robots: How Emotions Drive Self-Disclosure Behavior

    Guy Laban, Arvid Kappas, Val Morrison, Emily S. Cross. 2023 32nd IEEE International Conference on Robot and Human Interactive Communication (RO-MAN), 2023.

    TL;DR: When interacting with robots, people tend to share more personal information (self-disclosure) if the robot is perceived as comforting, or if the person is feeling negative emotions, like low mood, loneliness, and introversion.

    Self-disclosing to others can benefit emotional well-being, but socio-emotional barriers can limit people’s ability to do so. Self-disclosing towards social robots can help overcome these obstacles as robots lack judgment and can establish rapport. To further understand the influence of affective factors on people’s self-disclosure to social robots, this study examined the relationship between self-disclosure behaviour towards a social robot and people’s emotional states and their perception of the robot’s responses as comforting (i.e., being emphatic). The study included 1160 units of observation collected from 39 participants who conversed with the social robot Pepper (SoftBank Robotics) twice a week for 5 weeks (10 sessions in total), answering three personal questions in each session. Results show that perceiving the robot’s responses as more comforting was positively related to self-disclosure behaviour (in terms of disclosure duration in seconds, and disclosure length in number of words), and negative emotional states, such as lower mood, and higher feelings of loneliness and stress, were associated with higher rates of self-disclosure towards the robot. Additionally, higher rates of introversion significantly predicted higher rates of self-disclosure towards the robot. The study reveals the meaningful influence of affective states on how people behave when talking to social robots, especially when experiencing negative emotions. These findings may have implications for designing and developing social robots in therapeutic contexts.

  • Social robots as communication partners to support emotional well-being

    Guy Laban. PhD thesis, University of Glasgow, 2023.

    TL;DR: How people disclose their emotions to social robots, how this shapes the way they see the robots, and whether robots can support emotional well-being over time, including with informal caregivers.

    Interpersonal communication behaviors play a significant role in maintaining emotional well being. Self-disclosure is one such behavior that can have a meaningful impact on our emotional state. When we engage in self-disclosure, we can receive and provide support, improve our mood, and regulate our emotions. It also creates a comfortable space to share our feelings and emotions, which can have a positive impact on our overall mental and physical health. Social robots are gradually being introduced in a range of social and health settings. These autonomous machines can take on various forms and shapes and interact with humans using social behaviors and rules. They are being studied and introduced in psychosocial health interventions, including mental health and rehabilitation settings, to provide much- needed physical and social support to individuals. In my doctoral thesis, I aimed to explore how humans self-disclose and express their emotions to social robots and how this behavior can affect our perception of these agents. By studying speech-based communication interactions between humans and social robots, I wanted to investigate how social robots can support human emotional well-being. While social robots show great promise in offering social support, there are still many questions to consider before deploying them in actual care contexts. It is important to carefully evaluate their utility and scope in interpersonal communication settings, especially since social robots do not yet offer the same opportunities as humans for social interactions. My dissertation consists of three empirical chapters that investigate the underlying psychological mechanisms of perception and behaviour within human–robot communication and their potential deployment as interventions for emotional wellbeing. Chapter 1 offers a comprehensive introduction to the topic of emotional well-being and self-disclosure from a psychological perspective. I begin by providing an overview of the existing literature and theory in this field. Next, I delve into the social perception of social robots, presenting a theoretical framework to help readers understand how people view these machines. To illustrate this, I review some of the latest studies on social robots in care settings, as well as those exploring how robots can encourage people to self-disclose more about themselves. Finally, I explore the key concepts of self disclosure, including how it is defined, operationalized, and measured in experimental psychology and human–robot interaction research. In my first empirical chapter, Chapter 2, I explore how a social robot’s embodiment influences people’s disclosures in measurable terms, and how these disclosures differ from disclosures made to humans and disembodied agents. Chapter 3 studies how prolonged and intensive long-term interactions with a social robot affect people’s self-disclosure behavior towards the robot, perceptions of the robot, and how it affected factors related to well-being. Additionally, I examine the role of the interaction’s discussion theme. In Chapter 4, the final empirical chapter, I test a long-term and intensive social robot intervention with informal caregivers, people living with considerably difficult life situations. I investigate the potential of employing a social robot for eliciting self-disclosure among informal caregivers over time, supporting their emotional well-being, and implicitly encouraging them to adapt emotion regulation skills. In the final discussion chapter, Chapter 5, I summarise the current findings and discuss the contributions, implications and limitations of my work. I reflect on the contribution and challenges of this research approach and provide some future directions for researchers in the relevant fields. The results of these studies provide meaningful evidence for user experience, acceptance, and trust of social robots in different settings, including care, and demonstrate the unique psychological nature of these dynamic social interactions with social robots. Overall, this thesis contributes to the development of social robots that can support emotional well-being through self-disclosure interactions and provide insights into how social robots can be used as mental health interventions for individuals coping with emotional distress.

2022

  • Don’t Take it Personally: Resistance to Individually Targeted Recommendations from Conversational Recommender Agents

    Guy Laban, Theo Araujo. Proceedings of the 10th International Conference on Human-Agent Interaction, 2022.

    TL;DR: People are more likely to accept recommendations from AI chatbots if they ask for them directly, especially when the chatbot acts more human-like, as this reduces perceived risk and increases trust.

    Conversational recommender agents are artificially intelligent recommender systems that provide users with individually-tailored recommendations by targeting individual needs and communicating in a flowing dialogue. These are widely available online, communicating with users while demonstrating human-like (anthropomorphic) social cues. Nevertheless, little is known about the effect of their anthropomorphic cues on users’ resistance to the system and recommendations. Accordingly, this study examined the extent to which conversational recommender agents’ anthropomorphic cues and the type of recommendations provided (user-initiated and system-initiated) influenced users’ perceptions of control, trustworthiness, and the risk of using the platform. The study assessed how these perceptions, in turn, influence users’ adherence to the recommendations. An online experiment was conducted among users with conversational recommender agents and web recommender platforms that provided user-initiated or system-initiated restaurant recommendations. The results entail that user-initiated recommendations, compared to system-initiated, are less likely to affect users’ resistance to the system and are more likely to affect their adherence to the recommendations provided. Furthermore, the study’s findings suggest that these effects are amplified for conversational recommender agents, demonstrating anthropomorphic cues, in contrast to traditional systems as web recommender platforms.

  • User experience of human-robot long-term interactions

    Guy Laban, Arvid Kappas, Val Morrison, Emily S. Cross. Proceedings of the 10th International Conference on Human-Agent Interaction, 2022.

    TL;DR: This study found that people's interactions with a social robot improved over time, and they perceived the robot as getting better at communicating, regardless of the topics discussed.

    Since interactions with social robots are novel and exciting for many people, one particular concern in this specific area of human-robot interaction (HRI) is the extent to which human users will experience the interactions positively over time, when the robot’s novelty is particularly salient. In the current paper, we investigated users’ experience in long-term HRIs; how users perceive the ongoing interactions and the robot’s ability to sustain it over time. Therefore, here we examine the effect of the repeated measures (10 testing sessions) and the discussion theme (Covid-19 related vs general) on the way participants experienced the interaction quality with a social robot and perceived the robot’s communication competency over time. We found that despite individual differences between the participants, over time participants found the interactions with Pepper to be of higher quality and that Pepper’s communication skills got better. Nevertheless, our results also stressed that the discussion theme has no meaningful nor significant effect on the way people perceive Pepper and the interaction.

  • Social Robots as Communication Partners to Support Emotional Health and Well-Being

    Guy Laban. 2022 10th International Conference on Affective Computing and Intelligent Interaction Workshops and Demos (ACIIW), 2022.

    TL;DR: A doctoral consortium paper on how people disclose and express emotions to social robots, and how long-term conversations with robots can support well-being.

    Social robots are attracting increasing attention for their potential use in social and care settings. However, these agents do not (yet) offer the same opportunities as humans for social interactions. Given the importance of self-disclosure for psychological health, I am studying how humans disclose and express their emotions to social robots, and accordingly, how verbal interactions with social robots can support human users' emotional health and well-being. First, I was studying the role of embodiment in humans' disclosures to a social robot, compared to interactions with humans or disembodied agents. Following this, I studied the role of novelty in long-term human-robot interaction. Finally, I investigated how long-term speech-based interactions with social robots can be introduced as an intervention for individuals living in difficult situations. The results of the studies demonstrate the unique psychological nature of these dynamic social interactions with social robots.

  • Social robot advisors: effects of robot judgmental fallacies and context

    Torr Polakow, Guy Laban, Andrei Teodorescu, Jerome R. Busemeyer, Goren Gordon. Intell. Serv. Robot., 2022.

    TL;DR: Social robots can influence human decisions, even when they make mistakes, and this influence depends on the situation and the human's personality. Our study shows that people might trust a flawed robot more in some contexts than others.

    The role of social robots as advisors for decision-making is investigated. We examined how a robot advisor with logical reasoning and one with cognitive fallacies affected participants’ decision-making in different contexts. The participants were asked to make multiple decisions while receiving advice from both robots during the decision-making process. Participants had to choose which robot they agreed with and, at the end of the scenario, rank the possible options presented to them. After the interaction, participants were asked to assign jobs to the robots, e.g. jury or bartender. Based on the ‘like-me’ hypothesis and previous research of social mitigation of fallacious judgmental decisions, we have compared participants’ agreement with the two robots for each scenario to random choice using t-tests, as well as analysed the dynamical nature of the interaction, e.g. whether participants changed their choices based on the robots’ verbal opinion using Pearson correlations. Our results show that the robots had an effect on the participants’ responses, regardless of the robots’ fallaciousness, wherein participants changed their decisions based on the robot they agreed with more. Moreover, the context, presented as two different scenarios, also had an effect on the preferred robots, wherein an art auction scenario resulted in significantly increased agreement with the fallacious robot, whereas a detective scenario did not. Finally, an exploratory analysis showed that personality traits, e.g. agreeableness and neuroticism, and attitudes towards robots had an impact on which robot was assigned to these jobs. Taken together, the results presented here show that social robots’ effects on participants’ decision-making involve complex interactions between the context, the cognitive fallacies of the robot and the attitudes and personalities of the participants and should not be considered a single psychological construct.

  • Informal Caregivers Disclose Increasingly More to a Social Robot Over Time

    Guy Laban, Val Morrison, Arvid Kappas, Emily S. Cross. Extended Abstracts of the 2022 CHI Conference on Human Factors in Computing Systems, 2022.

    TL;DR: This study found that informal caregivers were more willing to open up and share personal information with a social robot over time, suggesting these robots could become valuable tools for supporting people under stress.

    Informal caregivers often struggle in managing to cope with both the stress and the practical demands of the caregiving situation. It has been suggested that digital solutions might be useful to monitor caregivers’ health and well-being, by providing early intervention and support. Given the importance of self-disclosure for psychological health, here we aimed to investigate the potential of employing a social robot for eliciting self-disclosure among informal caregivers over time. We conducted a longitudinal experiment across a five-week period, measuring participants’ disclosure duration (in seconds) and length (in number of words). Our preliminary results show a positive trend where informal caregivers speak for a longer time and share more information in their disclosures to a social robot across the five-week period. These results provide useful evidence supporting the potential deployment of social robots as intervention tools to help provide support for individuals suffering from stress and experiencing challenging life situations.

  • Ethics of Conversational User Interfaces

    Minha Lee, Jaisie Sin, Guy Laban, Matthias Kraus, Leigh Clark, Martin Porcheron, Benjamin R. Cowan, Asbjørn Følstad, Cosmin Munteanu, Heloisa Candello. CHI Conference on Human Factors in Computing Systems Extended Abstracts, 2022.

    TL;DR: A CHI 2022 workshop on the ethics of conversational user interfaces, centred on disclosure, transparency and vulnerable users.

    Building on the prior workshops on conversational user interfaces (CUIs) [2, 40], we tackle the topic of ethics of CUIs at CHI 2022. Though commercial CUI developments continue to rapidly advance, our scholarly dialogue on ethics of CUIs is underwhelming. The CUI community has implicitly been concerned with ethics, yet making it central to the growing body of work thus far has not been adequately done. Since ethics is a far-reaching topic, perspectives from philosophy, design, and engineering domains are integral to our CUI research community. For instance, philosophical traditions, e.g., deontology or virtue ethics, can guide ethical concepts that are relevant for CUIs, e.g., autonomy or trust. The practice of design through approaches like value sensitive design can inform how CUIs should be developed. Ethics comes into play with technical contributions, e.g., privacy-preserving data sharing between conversational systems. By considering such multidisciplinary angles, we come to a special topic of interest that ties together philosophy, design, and engineering: conversational disclosure, e.g., sharing personal information, transparency, e.g., as how to transparently convey relevant information in a conversational manner, and vulnerability of diverse user groups that should be taken into consideration.

  • Is Deep Learning a Valid Approach for Inferring Subjective Self-Disclosure in Human-Robot Interactions?

    Henry Powell, Guy Laban, Jean-Nöel George, Emily S. Cross. 2022 17th ACM/IEEE International Conference on Human-Robot Interaction (HRI), 2022.

    TL;DR: This paper investigates whether deep learning can accurately recognize varying levels of personal information shared with robots, finding that while it shows promise, current methods struggle with the nuances of human self-disclosure.

    One limitation of social robots has been the ability of the models they operate on to infer meaningful social information about people's subjective perceptions, specifically from non-invasive behavioral cues. Accordingly, our paper aims to demonstrate how different deep learning architectures trained on data from human-robot, human-human, and human-agent interactions can help artificial agents to extract meaning, in terms of people's subjective perceptions, in speech-based interactions. Here we focus on identifying people's perceptions of their subjective self-disclosure (i.e., to what extent one perceives to be sharing personal information with an agent). We approached this problem in a data-first manner, prioritizing high quality data over complex model architectures. In this context, we aimed to examine the extent to which relatively simple deep neural networks could extract non-lexical features related to this kind of subjective self perception. We show that five standard neural network architectures and one novel architecture, which we call a Hopfield Convolutional Neural Network, are all able to extract meaningful features from speech data relating to subjective self-disclosure.

  • Robo-Identity: Exploring Artificial Identity and Emotion via Speech Interactions

    Guy Laban, Sébastien Le Maguer, Minha Lee, Dimosthenis Kontogiorgos, Samantha Reig, Ilaria Torre, Ravi Tejwani, Matthew J. Dennis, Andre Pereira. 2022 17th ACM/IEEE International Conference on Human-Robot Interaction (HRI), 2022.

    TL;DR: The second Robo-Identity workshop, on how emotional synthetic speech shapes a robot's artificial identity and the ethical questions this raises.

    Following the success of the first edition of Robo-Identity, the second edition will provide an opportunity to expand the discussion about artificial identity. This year, we are focusing on emotions that are expressed through speech and voice. Synthetic voices of robots can resemble and are becoming indistinguishable from expressive human voices. This can be an opportunity and a constraint in expressing emotional speech that can (falsely) convey a human-like identity that can mislead people, leading to ethical issues. How should we envision an agent's artificial identity? In what ways should we have robots that maintain a machine-like stance, e.g., through robotic speech, and should emotional expressions that are increasingly human-like be seen as design opportunities? These are not mutually exclusive concerns. As this discussion needs to be conducted in a multidisciplinary manner, we welcome perspectives on challenges and opportunities from variety of fields. For this year's edition, the special theme will be “speech, emotion and artificial identity”.

  • Social Robots for Supporting Post-traumatic Stress Disorder Diagnosis and Treatment

    Guy Laban, Ziv Ben-Zion, Emily S. Cross. Frontiers in Psychiatry, 2022.

    TL;DR: This paper explores how social robots could help diagnose and treat PTSD, suggesting they might improve mental healthcare by making interventions more accessible and personalized, while also cautioning that careful research is needed.

    Post-Traumatic Stress Disorder (PTSD) is a severe psychiatric disorder with profound public health impact due to its high prevalence, chronic nature, accompanying functional impairment, and frequently occurring comorbidities. Early PTSD symptoms, often observed shortly after trauma exposure, abate with time in the majority of those who initially express them, yet leave a significant minority with chronic PTSD. While the past several decades of PTSD research have produced substantial knowledge regarding the mechanisms and consequences of this debilitating disorder, the diagnosis of and available treatments for PTSD still face significant challenges. Here, we discuss how novel therapeutic interventions involving social robots can potentially offer meaningful opportunities for overcoming some of the present challenges. As the application of social robotics-based interventions in the treatment of mental disorders is only in its infancy, it is vital that careful, well-controlled research is conducted to evaluate their efficacy, safety, and ethics. Nevertheless, we are hopeful that robotics-based solutions could advance the quality, availability, specificity and scalability of care for PTSD.

2021

  • Perceptions of anthropomorphism in a chatbot dialogue: The role of animacy and intelligence

    Guy Laban. HAI '21: Proceedings of the 9th International Conference on Human-Agent Interaction, 2021.

    TL;DR: People see chatbots as more "alive" (animate) than regular websites, which makes them feel more human-like, but chatbots aren't necessarily perceived as smarter.

    When interacting with embodied agents, users often rely on a variety of cues from the agent's embodiment to form perceptions of animacy and intelligence, including its appearance and behaviour. Due to chatbots' disembodiment, users' perceptions of a chatbot's animacy and intelligence are mostly dependent on the textual properties of the dialogue. The current study aims to investigate the mediating role of perceptions of chatbot's intelligence and animacy on users' perceptions of the chatbot's anthropomorphism. An online experiment was conducted with a chatbot and a web platform. Both systems asked users three basic questions for providing a restaurant recommendation. By communicating the same content via different modalities of communication (i.e., flowing dialogue and traditional web interface); this study compares the differences in these perceptions between chatbots to traditional web platforms. The results of a mediation analysis entail that the chatbot was perceived as more animate than the web platform, and accordingly, it was perceived as more anthropomorphic than the web platform as users' perceptions of animacy fully mediated this effect. Also, there is no evidence for differences in users' perceptions of intelligence between the chatbot and the web platform.

  • Future directions for chatbot research: an interdisciplinary research agenda

    Asbjørn Følstad, Theo Araujo, Effie Lai-Chong Law, Petter Bae Brandtzaeg, Symeon Papadopoulos, Lea Reis, Marcos Baez, Guy Laban, Patrick McAllister, Carolin Ischen, Rebecca Wald, Fabio Catania, Raphael Meyer von Wolff, Sebastian Hobert, Ewa Luger. Computing, 2021.

    TL;DR: A five-year research agenda for chatbots across six themes, from users and design to collaboration, democratisation, ethics and privacy.

    Chatbots are increasingly becoming important gateways to digital services and information—taken up within domains such as customer service, health, education, and work support. However, there is only limited knowledge concerning the impact of chatbots at the individual, group, and societal level. Furthermore, a number of challenges remain to be resolved before the potential of chatbots can be fully realized. In response, chatbots have emerged as a substantial research area in recent years. To help advance knowledge in this emerging research area, we propose a research agenda in the form of future directions and challenges to be addressed by chatbot research. This proposal consolidates years of discussions at the CONVERSATIONS workshop series on chatbot research. Following a deliberative research analysis process among the workshop participants, we explore future directions within six topics of interest: (a) users and implications, (b) user experience and design, (c) frameworks and platforms, (d) chatbots for collaboration, (e) democratizing chatbots, and (f) ethics and privacy. For each of these topics, we provide a brief overview of the state of the art, discuss key research challenges, and suggest promising directions for future research. The six topics are detailed with a 5-year perspective in mind and are to be considered items of an interdisciplinary research agenda produced collaboratively by avid researchers in the field.

  • What Makes a Robot Social? A Review of Social Robots from Science Fiction to a Home or Hospital Near You

    Anna Henschel, Guy Laban, Emily S. Cross. Current Robotics Reports, 2021.

    TL;DR: This paper explores what defines a "social robot" by looking at how they're portrayed in fiction and used in real-world applications, showing their potential to interact and help people in various settings.

    Purpose of Review We provide an outlook on the definitions, laboratory research, and applications of social robots, with an aim to understand what makes a robot social-in the eyes of science and the general public. Recent Findings Social robots demonstrate their potential when deployed within contexts appropriate to their form and functions. Some examples include companions for the elderly and cognitively impaired individuals, robots within educational settings, and as tools to support cognitive and behavioural change interventions. Summary Science fiction has inspired us to conceive of a future with autonomous robots helping with every aspect of our daily lives, although the robots we are familiar with through film and literature remain a vision of the distant future. While there are still miles to go before robots become a regular feature within our social spaces, rapid progress in social robotics research, aided by the social sciences, is helping to move us closer to this reality.

  • Tell me more! Assessing interactions with social robots from speech

    Guy Laban, Jean-Noël George, Val Morrison, Emily S. Cross. Paladyn, Journal of Behavioral Robotics, 2021.

    TL;DR: This paper shows how analyzing the speech patterns of people interacting with robots can reveal how they view the interaction and how they express themselves.

    As social robots are increasingly introduced into health interventions, one potential area where they might prove valuable is in supporting people’s psychological health through conversation. Given the importance of self-disclosure for psychological health, this study assessed the viability of using social robots for eliciting rich disclosures that identify needs and emotional states in human interaction partners. Three within-subject experiments were conducted with participants interacting with another person, a humanoid social robot, and a disembodied conversational agent (voice assistant). We performed a number of objective evaluations of disclosures to these three agents via speech content and voice analyses and also probed participants’ subjective evaluations of their disclosures to three agents. Our findings suggest that participants overall disclose more to humans than artificial agents, that agents’ embodiment influences disclosure quantity and quality, and that people are generally aware of differences in their personal disclosures to three agents studied here. Together, the findings set the stage for further investigation into the psychological underpinnings of self-disclosures to artificial agents and their potential role in eliciting disclosures as part of mental and physical health interventions.

2020

  • The Effect of Personalization Techniques in Users' Perceptions of Conversational Recommender Systems

    Guy Laban, Theo Araujo. Proceedings of the 20th ACM International Conference on Intelligent Virtual Agents, 2020.

    TL;DR: In conversational recommendation systems, users feel less in control and perceive more risk when the system initiates recommendations compared to when users initiate them, but the way recommendations are started doesn't change how human-like users find the system.

    Conversational recommender systems provide users with individually tailored recommendations in a flowing dialogue. These require users to disclose information proactively or reactively for receiving personalized recommendations, which can trigger users' resistance to the platform and to the recommendations. Accordingly, this study examined the extent to which user-initiated and system-initiated recommendations provided by a conversational recommender system influenced users' perceptions of it. The results of an online experiment entail that when recommendations are system-initiated, as compared to user-initiated, users perceive to be in less control and perceive the system as riskier. Furthermore, the results stress that systems that provide user-initiated or system-initiated recommendations do not differ in users' perceptions of anthropomorphism.

  • Binge-watching (Netflix) product placement: A content analysis on different product placements in Netflix originals vs. non-Netflix originals, and drama vs. comedy shows

    Guy Laban, Chamoetal Zeidler, Eline Brussee. MediArXiv, 2020.

    TL;DR: Netflix originals carry more product placement than other shows on the platform, and dramas more than comedies.

    The popularity of digital streaming platforms, such as Netflix, is gradually growing. While digital streaming platforms share similar economic interests as traditional producing and broadcasting companies, the two differ in operation and nature of consumption. These platforms constructed new means of consuming content, offering licensed and original content. This study describes and examines how product placements, a popular advertising strategy that integrates commercials in media entertainment content, are applied in an up-to-date digital environment. A content analysis was conducted of product placements in comedy and drama shows within Netflix originals and non-Netflix originals shows. The study results indicate that Netflix original shows include more product placements than non-Netflix original shows. Moreover, drama shows include more product placements than comedy shows, and these are more explicit and visual. Finally, it was found that product placements categories can be predicted by the genre of the show, but not by the production type.

  • Let's Talk About It! Subjective and Objective Disclosures to Social Robots

    Guy Laban, Val Morrison, Emily S. Cross. Companion of the 2020 ACM/IEEE International Conference on Human-Robot Interaction, 2020.

    TL;DR: People felt they disclosed more to a human than to a robot or a conversational agent, yet what they actually said was much the same across all three.

    This study aims to test the viability of using social robots for eliciting rich disclosures from humans to identify their needs and emotional states. Self-disclosure has been studied in the psychological literature in many ways, addressing both peoples' subjective perceptions of their disclosures, as well as objective disclosures evaluating these via direct observation and analysis of verbal and written output. Here we are interested in how people disclose (non-sensitive) personal information to robots, in an aim to further understand the differences between one's subjective perceptions of disclosure compared to evidence of disclosure from the shared content. An experimental design is suggested for evaluating disclosure to social robots compared to humans and conversational agents. Initial results suggest that while people perceive they disclose more to humans than to humanoid social robots or conversational agents, no actual observed differences in the content of the disclosure emerges between the three agents.

  • Working Together with Conversational Agents: The Relationship of Perceived Cooperation with Service Performance Evaluations

    Guy Laban, Theo Araujo. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2020.

    TL;DR: This paper explores how consumers' perceptions of working together with an AI chatbot impact their satisfaction with the service they receive. It finds that feeling cooperative with the chatbot leads to better service evaluations.

    Conversational agents are gradually being deployed by organizations in service settings to communicate with and solve problems together with consumers. The current study investigates how consumers’ perceptions of cooperation with conversational agents in a service context are associated with their perceptions about agents’ anthropomorphism, social presence, the quality of the information provided by an agent, and the agent service performance. An online experiment was conducted in which participants performed a service-oriented task with the assistance of conversational agents developed specifically for the study and evaluated the performance and attributes of the agents. The results suggest a direct positive link between perceiving a conversational agent as cooperative and perceiving it to be more anthropomorphic, with higher levels of social presence and providing better information quality. Moreover, the results also show that the link between perceiving an agent as cooperative and the agent’s service performance is mediated by perceptions of the agent’s anthropomorphic cues and the quality of the information provided by the agent.