arXiv subjects

Lize Alberts

Publications and source records attributed to Lize Alberts.

6 recordsLinked to original sources

arXiv2024

CURATe: Benchmarking Personalised Alignment of Conversational AI Assistants

We introduce a multi-turn benchmark for evaluating personalised alignment in LLM-based AI assistants, focusing on their ability to handle user-provided safety-critical contexts. Our assessment of ten leading models across five scenarios (with 337 use cases each) reveals systematic inconsistencies in maintaining user-specific consideration, with even top-rated "harmless" models making recommendations that should be recognised as obviously harmful to the user given the context provided. Key failure modes include inappropriate weighing of conflicting preferences, sycophancy (prioritising desires above safety), a lack of attentiveness to critical user information within the context window, and inconsistent application of user-specific knowledge. The same systematic biases were observed in OpenAI's o1, suggesting that strong reasoning capacities do not necessarily transfer to this kind of personalised thinking. We find that prompting LLMs to consider safety-critical context significantly improves performance, unlike a generic 'harmless and helpful' instruction. Based on these findings, we propose research directions for embedding self-reflection capabilities, online user modelling, and dynamic risk assessment in AI assistants. Our work emphasises the need for nuanced, context-aware approaches to alignment in systems designed for persistent human interaction, aiding the development of safe and considerate AI assistants.

cs.HC↗

arXiv2024

The Ethics of Advanced AI Assistants

This paper focuses on the opportunities and the ethical and societal risks posed by advanced AI assistants. We define advanced AI assistants as artificial agents with natural language interfaces, whose function is to plan and execute sequences of actions on behalf of a user, across one or more domains, in line with the user's expectations. The paper starts by considering the technology itself, providing an overview of AI assistants, their technical foundations and potential range of applications. It then explores questions around AI value alignment, well-being, safety and malicious uses. Extending the circle of inquiry further, we next consider the relationship between advanced AI assistants and individual users in more detail, exploring topics such as manipulation and persuasion, anthropomorphism, appropriate relationships, trust and privacy. With this analysis in place, we consider the deployment of advanced assistants at a societal scale, focusing on cooperation, equity and access, misinformation, economic impact, the environment and how best to evaluate advanced AI assistants. Finally, we conclude by providing a range of recommendations for researchers, developers, policymakers and public stakeholders.

cs.CY↗

arXiv2024

Designing for Sustained Motivation: A Review of Self-Determination Theory in Behaviour Change Technologies

Recent years have seen a surge in applications and technologies aimed at motivating users to achieve personal goals and improve their wellbeing. However, these often fail to promote long-term behaviour change, and sometimes even backfire. We consider how self-determination theory (SDT), a metatheory of human motivation and wellbeing, can help explain why such technologies fail, and how they may better help users internalise the motivation behind their goals and make enduring changes in their behaviour. In this work, we systematically reviewed 15 papers in the ACM Digital Library that apply SDT to the design of behaviour change technologies (BCTs). We identified 50 suggestions for design features in BCTs, grounded in SDT, that researchers have applied to enhance user motivation. However, we find that SDT is often leveraged to optimise engagement with the technology itself rather than with the targeted behaviour change per se. When interpreted through the lens of SDT, the implication is that BCTs may fail to cultivate sustained changes in behaviour, as users' motivation depends on their enjoyment of the intervention, which may wane over time. An underexplored opportunity remains for designers to leverage SDT to support users to internalise the ultimate goals and value of certain behaviour changes, enhancing their motivation to sustain these changes in the long term.

cs.HC↗

arXiv2024

Should agentic conversational AI change how we think about ethics? Characterising an interactional ethics centred on respect

With the growing popularity of conversational agents based on large language models (LLMs), we need to ensure their behaviour is ethical and appropriate. Work in this area largely centres around the 'HHH' criteria: making outputs more helpful and honest, and avoiding harmful (biased, toxic, or inaccurate) statements. Whilst this semantic focus is useful when viewing LLM agents as mere mediums or output-generating systems, it fails to account for pragmatic factors that can make the same speech act seem more or less tactless or inconsiderate in different social situations. With the push towards agentic AI, wherein systems become increasingly proactive in chasing goals and performing actions in the world, considering the pragmatics of interaction becomes essential. We propose an interactional approach to ethics that is centred on relational and situational factors. We explore what it means for a system, as a social actor, to treat an individual respectfully in a (series of) interaction(s). Our work anticipates a set of largely unexplored risks at the level of situated social interaction, and offers practical suggestions to help agentic LLM technologies treat people well.

cs.CL↗

arXiv2023

Computers as Bad Social Actors: Dark Patterns and Anti-Patterns in Interfaces that Act Socially

Technologies increasingly mimic human-like social behaviours. Beyond prototypical conversational agents like chatbots, this also applies to basic automated systems like app notifications or self-checkout machines that address or 'talk to' users in everyday situations. Whilst early evidence suggests social cues may enhance user experience, we lack a good understanding of when, and why, their use may be inappropriate. Building on a survey of English-speaking smartphone users (n=80), we conducted experience sampling, interview, and workshop studies (n=11) to elicit people's attitudes and preferences regarding how automated systems talk to them. We thematically analysed examples of phrasings/conduct participants disliked, the reasons they gave, and what they would prefer instead. One category of inappropriate behaviour we identified regards the use of social cues as tools for manipulation. We describe four unwanted tactics interfaces use: agents playing on users' emotions (e.g., guilt-tripping or coaxing them), being pushy, `mothering' users, or being passive-aggressive. Another category regards pragmatics: personal or situational factors that can make a seemingly friendly or helpful utterance come across as rude, tactless, or invasive. These include failing to account for relevant contextual particulars (e.g., embarrassing users in public); expressing obviously false personalised care; or treating a user in ways that they find inappropriate for the system's role or the nature of their relationship. We discuss these behaviours in terms of an emerging 'social' class of dark and anti-patterns. Drawing from participant recommendations, we offer suggestions for improving how interfaces treat people in interactions, including broader normative reflections on treating users respectfully.

cs.HC↗

arXiv2022

Not Cheating on the Turing Test: Towards Grounded Language Learning in Artificial Intelligence

Recent hype surrounding the increasing sophistication of language processing models has renewed optimism regarding machines achieving a human-like command of natural language. Research in the area of natural language understanding (NLU) in artificial intelligence claims to have been making great strides in this area, however, the lack of conceptual clarity/consistency in how 'understanding' is used in this and other disciplines makes it difficult to discern how close we actually are. In this interdisciplinary research thesis, I integrate insights from cognitive science/psychology, philosophy of mind, and cognitive linguistics, and evaluate it against a critical review of current approaches in NLU to explore the basic requirements--and remaining challenges--for developing artificially intelligent systems with human-like capacities for language use and comprehension.

cs.CL↗