We have deployed an LLM-based spoken dialogue system in a real hospital. The ARI social robot embodies our system, which patients and their companions can have multi-party conversations with together. In order to enable this multi-party ability, multimodality is critical. Our system, therefore, receives speech and video as input, and generates both speech and gestures (arm, head, and eye movements). In this paper, we describe our complex setting and the architecture of our dialogue system. Each component is detailed, and a video of the full system is available with the appropriate components highlighted in real-time. Our system decides when it should take its turn, generates human-like clarification requests when the patient pauses mid-utterance, answers in-domain questions (grounding to the in-prompt knowledge), and responds appropriately to out-of-domain requests (like generating jokes or quizzes). This latter feature is particularly remarkable as real patients often utter unexpected sentences that could not be handled previously.

When deploying LLMs in certain commercial or research settings, domain specific knowledge must be explicitly provided within the prompt. This in-prompt knowledge can conflict with an LLM’s static world knowledge learned at pre-training, causing model hallucination (see examples in Table 1). In safety-critical settings, like healthcare and finance, these hallucinations can harm vulnerable users. We have curated a QA corpus containing information that LLMs could not have seen at pre-training. Using our corpus, we have probed various LLMs, manipulating both the prompt and the knowledge representation. We have found that our ‘Jodie’ prompt consistently improves the model’s textual grounding to the given knowledge, and in-turn the overall answer accuracy. This is true in both the healthcare and finance domains - improving accuracy by up to 28% (mean: 12%). We have also identified that hierarchical and direct node-property graph structures could lead to more interpretable and controllable systems that provide a natural language interface with real-time in-domain knowledge. Our corpus will enable further work on this critical challenge.

People understand and produce language incrementally on a word by word basis. This gives rise to many characteristic conversational phenomena including long mid-sentence pauses that are followed by incremental clarification requests (iCRs) intended to recover the rest of the truncated turn (see Fig. 1; (A), (B), (C)). The ability to generate iCRs is important in natural conversational AI systems, and crucial to their accessibility to users with memory impairment. In this paper, we collect, release and analyse SLUICE-CR: a large corpus of 3000 human produced iCRs. We then use this corpus to probe the incremental processing capability of a number of state of the art LLMs by evaluating the quality of the model’s generated iCRs in response to incomplete questions. Our evaluations show that the ability to generate contextually appropriate iCRs only emerges at larger LLM sizes, and only when prompted with example iCRs from our corpus. They also indicate that autoregressive LMs are, in principle, able to both understand and generate language incrementally.

This paper evaluates the extent to which current LLMs can capture task-oriented multi-party conversations (MPCs). We have recorded and transcribed 29 MPCs between patients, their companions, and a social robot in a hospital. We then annotated this corpus for multi-party goal-tracking and intent-slot recognition. People share goals, answer each other’s goals, and provide other people’s goals in MPCs - none of which occur in dyadic interactions. To understand user goals in MPCs, we compared three methods in zero-shot and few-shot settings: we fine-tuned T5, created pre-training tasks to train DialogLM using LED, and employed prompt engineering techniques with GPT-3.5-turbo, to determine which approach can complete this novel task with limited data. GPT-3.5-turbo significantly outperformed the others in a few-shot setting. The ‘reasoning’ style prompt, when given 7% of the corpus as example annotated conversations, was the best performing method. It correctly annotated 62.32% of the goal tracking MPCs, and 69.57% of the intent-slot recognition MPCs. A ‘story’ style prompt increased model hallucination, which could be detrimental if deployed in safety-critical settings. We conclude that multi-party conversations still challenge state-of-the-art LLMs.

Speech production is nuanced and unique to every individual, but today’s Spoken Dialogue Systems (SDSs) are trained to use general speech patterns to successfully improve performance on various evaluation metrics. However, these patterns do not apply to certain user groups - often the very people that can benefit the most from SDSs. For example, people with dementia produce more disfluent speech than the general population. The healthcare domain is now a popular setting for spoken dialogue and human-robot interaction research. This trend is similar when observing company behaviour. Charities promote industry voice assistants, the creators are getting HIPAA compliance, and their features sometimes target vulnerable user groups. It is therefore critical to adapt SDSs to be more accessible.

Socially Assistive Robots (SARs) have the potential to play an increasingly important role in a variety of contexts including healthcare, but most existing systems have very limited interactive capabilities. We will demonstrate a robot receptionist that not only supports task-based and social dialogue via natural spoken conversation but is also capable of visually grounded dialogue; able to perceive and discuss the shared physical environment (e.g. helping users to locate personal belongings or objects of interest). Task-based dialogues include check-in, navigation and FAQs about facilities, alongside social features such as chit-chat, access to the latest news and a quiz game to play while waiting. We also show how visual context (objects and their spatial relations) can be combined with linguistic representations of dialogue context, to support visual dialogue and question answering. We will demonstrate the system on a humanoid ARI robot, which is being deployed in a hospital reception area.

Speech production is nuanced and unique to every individual, but today’s Spoken Dialogue Systems (SDSs) are trained to use general speech patterns to successfully improve performance on various evaluation metrics. However, these patterns do not apply to certain user groups - often the very people that can benefit the most from SDSs. For example, people with dementia produce more disfluent speech than the general population. In order to evaluate systems with specific user groups in mind, and to guide the design of such systems to deliver maximum benefit to these users, data must be collected securely. In this short paper we present CVR-SI, a bespoke tool for ethical data collection. Designed for the healthcare domain, we argue that it should also be used in more general settings. We detail how off-the-shelf solutions fail to ensure that sensitive data remains secure and private. We then describe the ethical design and security features of our device, with a full guide on how to build both the hardware and software components of CVR-SI. Our design ensures inclusivity to all researchers in this field, particularly those who are not hardware experts. This guarantees everyone can collect appropriate data for human evaluation ethically, securely, and in a timely manner.

The next generation of conversational AI systems need to: (1) process language incrementally, token-by-token to be more responsive and enable handling of conversational phenomena such as pauses, restarts and self-corrections; (2) reason incrementally allowing meaning to be established beyond what is said; (3) be transparent and controllable, allowing designers as well as the system itself to easily establish reasons for particular behaviour and tailor to particular user groups, or domains. In this short paper we present ongoing preliminary work combining Dynamic Syntax (DS) - an incremental, semantic grammar framework - with the Resource Description Framework (RDF). This paves the way for the creation of incremental semantic parsers that progressively output semantic RDF graphs as an utterance unfolds in real-time. We also outline how the parser can be integrated with an incremental reasoning engine through RDF. We argue that this DS-RDF hybrid satisfies the desiderata listed above, yielding semantic infrastructure that can be used to build responsive, real-time, interpretable Conversational AI that can be rapidly customised for specific user groups such as people with dementia.

Visual Question Answering (VQA) systems are increasingly adept at a variety of tasks, and this technology can be used to assist blind and partially sighted people. To do this, the system’s responses must not only be accurate, but usable. It is also vital for assistive technologies to be designed with a focus on: (1) privacy, as the camera may capture a user’s mail, medication bottles, or other sensitive information; (2) transparency, so that the system’s behaviour can be explained and trusted by users; and (3) controllability, to tailor the system for a particular domain or user group. We have therefore extended a conversational VQA framework, called Aye-saac, with these objectives in mind. Specifically, we gave Aye-saac the ability to answer visual questions in the kitchen, a particularly challenging area for visually impaired people. Our system can now answer questions about quantity, positioning, and system confidence in regards to 299 kitchen objects. Questions about the spatial relations between these objects are particularly helpful to visually impaired people, and our system output more usable answers than other state of the art end-to-end VQA systems.

Automatic Speech Recognition (ASR) systems are increasingly powerful and more accurate, but also more numerous with several options existing currently as a service (e.g. Google, IBM, and Microsoft). Currently the most stringent standards for such systems are set within the context of their use in, and for, Conversational AI technology. These systems are expected to operate incrementally in real-time, be responsive, stable, and robust to the pervasive yet peculiar characteristics of conversational speech such as disfluencies and overlaps. In this paper we evaluate the most popular of such systems with metrics and experiments designed with these standards in mind. We also evaluate the speaker diarization (SD) capabilities of the same systems which will be particularly important for dialogue systems designed to handle multi-party interaction. We found that Microsoft has the leading incremental ASR system which preserves disfluent materials and IBM has the leading incremental SD system in addition to the ASR that is most robust to speech overlaps. Google strikes a balance between the two but none of these systems are yet suitable to reliably handle natural spontaneous conversations in real-time.