arXiv subjects

Gray Manicom

Publications and source records attributed to Gray Manicom.

3 recordsLinked to original sources

arXiv2026

The CRAFT principles for the responsible use of large language models in policymaking

Policymakers around the world face the question of how to use artificial intelligence in general, and large language models in particular, to improve the policymaking process. Used well, large language models can strengthen the collection, interpretation and synthesis of policy-relevant information and the drafting of policy-relevant output. Yet the use of large language models in policymaking is associated with risks. Output that is plausible but not necessarily correct, bias resulting from unrepresentative training data, the exposure of sensitive information and, over time, deskilling and dependency can erode trust if large language models are not used thoughtfully. The CRAFT principles - control, rigour, accountability, fairness and transparency - offer a way to make the most of large language models in policymaking while managing the risks.

cs.CY↗

arXiv2026

User identity conditions moral wrongness ratings in non-reasoning large language models

This study adopts a behavioural bottom-up approach to AI value alignment to investigate whether an implicitly conveyed user identity shifts the moral evaluations of large language models (LLMs). Through a structured, multi-turn conversational protocol across 12,000 interactions, we evaluate AI value alignment in two non-reasoning models, gpt-4.1-mini-2025-04-14 and gemini-2.5-flash-lite. Rather than instructing the models to adopt a persona or prompting them with explicit moral stances, the user's professional role is introduced purely through value-neutral reasoning. The models are then asked for wrongness ratings from 0-100 on ten common-morality rules from Gert's moral framework. The results show that moral judgments vary with the user's role across both models. While grave-harm acts like killing exhibit a strong ceiling effect, contestable rule-governed acts demonstrate role-conditioned shifts that mirror the relationship between the user's profession and the act being rated. These findings demonstrate that unintended contextual conditioning via user identity permeates LLM moral evaluations, posing questions for the AI value alignment discourse regarding how to define acceptable bounds for role-based moral divergence. By doing so, the results contribute to reframing the AI value alignment discourse by suggesting future research on dynamic moral bounds rather than static moral principles or rules as frame of reference.

cs.CY↗

arXiv2025

Modelling Immunity in Agent-based Models

Vaccination policies play a central role in public health interventions and models are often used to assess the effectiveness of these policies. Many vaccines are leaky, in which case the observed vaccine effectiveness depends on the force of infection. Within models, the immunity parameters required for agent-based models to achieve observed vaccine effectiveness values are further influenced by model features such as its transmission algorithm, contact network structure, and approach to simulating vaccination. We present a method for determining parameters in agent-based models such that a set of target immunity values is achieved. We construct a dataset of desired population-level immunity values against various disease outcomes considering both vaccination and prior infection from COVID-19. This dataset incorporates immunological data, data collection methodologies, immunity models, and biological insights. We then describe how we choose minimal parameters for continuous waning immunity curves that result in those target values being realized in simulations. We use simulations of the household secondary attack rates to establish a relationship between the protection per infection attempt and overall immunity, thus accounting for the dependence of protection from acquisition on model features and the force of infection.

q-bio.PE↗