Abstract In Formula One, often described as the pinnacle of motorsport, teams compete to design and produce the fastest cars, driven by some of the best drivers in the world, in order to win races. However, a team has little chance of success without effective race strategy , i.e. selecting which tyre compounds to use and when to take pitstops to change between them. Teams’ methods for solving this problem are usually limited to linear optimisation, while some run Monte Carlo simulations in simple, best-case situations; these approaches thus fail to take into account the complex interactions between teams’ strategies and tactics in this unpredictable multi-agent environment. Further, there is low uptake of AI in this domain, potentially due to a lack of trust in these “black-box” models. In this work, we enable the massive potential of reinforcement learning (RL) models in this space using post-hoc techniques from explainable AI. Specifically, we introduce Race Strategy Reinforcement Learning (RSRL), an RL model which allows us to control the strategies of cars in race simulations, with explanations for their actions to help foster trust in users. We first demonstrate that RSRL outperforms baselines of hard-coded and Monte-Carlo strategies, presenting opportunities for improving race strategy for all Formula One teams, and potentially beyond, especially in other areas of motorsport. Next, we analyse RSRL’s generalisability to unseen tracks and show how performance on one or multiple tracks can be prioritised via training. We then exhibit the fidelity and comprehensibility of the deployed explanations towards improving user trust in RSRL’s decisions. Finally, we highlight the emergent tactics , i.e. emergent behaviours representing real-world tactics, learnt by RSRL, pointing towards the general applicability of RL for modelling and even influencing race strategy in Formula One.

Soft robotic devices, known for their high compliance, are increasingly being used in assistance and rehabilitation. However, the limited force output of soft actuators has hindered their broader adoption. In this study, a lobster-tail-inspired high-force-output soft pneumatic bending actuator (SPBA) is developed, featuring a soft deformable body and a rigid kirigami limiting shell. The SPBA, with a radius of 10 mm, can generate forces of approximately 22 N at an internal pressure of 0.1 MPa and 36.43 N at 0.16 MPa. An analytical model based on the Euler–Bernoulli beam theory, incorporating a hyperelastic material model, has been constructed to predict the deformation and force of the actuated SPBA. This model demonstrates good agreement with simulated and experimental results. For assistance, a soft robotic gripper with four SPBAs can lift a weight of 5.38 kg at 0.26 MPa. For rehabilitation, an SPBA-based hand exoskeleton has been developed, demonstrating significant effectiveness in mitigating hand spasticity following strokes. This study introduces a novel SPBA design with promising potential for future applications in grasping, assistance, and rehabilitation.

Reward models (RMs) are a crucial component in the alignment of large language models’ (LLMs) outputs with human values. RMs approximate human preferences over possible LLM responses to the same prompt by predicting and comparing reward scores. However, as they are typically modified versions of LLMs with scalar output heads, RMs are large black boxes whose predictions are not explainable. More transparent RMs would enable improved trust in the alignment of LLMs. In this work, we propose to use contrastive explanations to explain any binary response comparison made by an RM. Specifically, we generate a diverse set of new comparisons similar to the original one to characterise the RM’s local behaviour. The perturbed responses forming the new comparisons are generated to explicitly modify manually specified high-level evaluation attributes, on which analyses of RM behaviour are grounded. In quantitative experiments, we validate the effectiveness of our method for finding high-quality contrastive explanations. We then showcase the qualitative usefulness of our method for investigating global sensitivity of RMs to each evaluation attribute, and demonstrate how representative examples can be automatically extracted to explain and compare behaviours of different RMs. We see our method as a flexible framework for RM explanation, providing a basis for more interpretable and trustworthy LLM alignment.

The increasing use of Machine Learning (ML) models to aid decision-making in high-stakes industries demands explainability to facilitate trust. Counterfactual Explanations (CEs) are ideally suited for this, as they can offer insights into the predictions of an ML model by illustrating how changes in its input data may lead to different outcomes. However, for CEs to realise their explanatory potential, significant challenges remain in ensuring their robustness under slight changes in the scenario being explained. Despite the widespread recognition of CEs' robustness as a fundamental requirement, a lack of standardised tools and benchmarks hinders a comprehensive and effective comparison of robust CE generation methods. In this paper, we introduce RobustX, an open-source Python library implementing a collection of CE generation and evaluation methods, with a focus on the robustness property. RobustX provides interfaces to several existing methods from the literature, enabling streamlined access to state-of-the-art techniques. The library is also easily extensible, allowing fast prototyping of novel robust CE generation and evaluation methods.

Test-time scaling improves large language models' (LLMs) performance by allocating more compute budget during inference. To achieve this, existing methods often require intricate modifications to prompting and sampling strategies. In this work, we introduce representation consistency (RC), a test-time scaling method for aggregating answers drawn from multiple candidate responses of an LLM regardless of how they were generated, including variations in prompt phrasing and sampling strategy. RC enhances answer aggregation by not only considering the number of occurrences of each answer in the candidate response set, but also the consistency of the model's internal activations while generating the set of responses leading to each answer. These activations can be either dense (raw model activations) or sparse (encoded via pretrained sparse autoencoders). Our rationale is that if the model's representations of multiple responses converging on the same answer are highly variable, this answer is more likely to be the result of incoherent reasoning and should be down-weighted during aggregation. Importantly, our method only uses cached activations and lightweight similarity computations and requires no additional model queries. Through experiments with four open-source LLMs and four reasoning datasets, we validate the effectiveness of RC for improving task performance during inference, with consistent accuracy improvements (up to 4\%) over strong test-time scaling baselines. We also show that consistency in the sparse activation signals aligns well with the common notion of coherent reasoning.

AI has become pervasive in recent years, but state-of-the-art approaches predominantly neglect the need for AI systems to be contestable. Instead, contestability is advocated by AI guidelines (e.g. by the OECD) and regulation of automated decision-making (e.g. GDPR). In this position paper we explore how contestability can be achieved computationally in and for AI. We argue that contestable AI requires dynamic (human-machine and/or machine-machine) explainability and decision-making processes, whereby machines can 1. interact with humans and/or other machines to progressively explain their outputs and/or their reasoning as well as assess grounds for contestation provided by these humans and/or other machines, and 2. revise their decision-making processes to redress any issues successfully raised during contestation. Given that much of the current AI landscape is tailored to static AIs, the need to accommodate contestability will require a radical rethinking, that, we argue, computational argumentation is ideally suited to support.

Counterfactual Explanations (CEs) have emerged as a major paradigm in explainable AI research, providing recourse recommendations for users affected by the decisions of machine learning models. However, CEs found by existing methods often become invalid when slight changes occur in the parameters of the model they were generated for. The literature lacks a way to provide exhaustive robustness guarantees for CEs under model changes, in that existing methods to improve CEs' robustness are mostly heuristic, and the robustness performances are evaluated empirically using only a limited number of retrained models. To bridge this gap, we propose a novel interval abstraction technique for parametric machine learning models, which allows us to obtain provable robustness guarantees for CEs under a possibly infinite set of plausible model changes Δ. Based on this idea, we formalise a robustness notion for CEs, which we call Δ-robustness, in both binary and multi-class classification settings. We present procedures to verify Δ-robustness based on Mixed Integer Linear Programming, using which we further propose algorithms to generate CEs that are Δ-robust. In an extensive empirical study involving neural networks and logistic regression models, we demonstrate the practical applicability of our approach. We discuss two strategies for determining the appropriate hyperparameters in our method, and we quantitatively benchmark CEs generated by eleven methods, highlighting the effectiveness of our algorithms in finding robust CEs.

The use of counterfactual explanations (CFXs) is an increasingly popular explanation strategy for machine learning models. However, recent studies have shown that these explanations may not be robust to changes in the underlying model (e.g., following retraining), which raises questions about their reliability in real-world applications. Existing attempts towards solving this problem are heuristic, and the robustness to model changes of the resulting CFXs is evaluated with only a small number of retrained models, failing to provide exhaustive guarantees. To remedy this, we propose ∆-robustness, the first notion to formally and deterministically assess the robustness (to model changes) of CFXs for neural networks. We introduce an abstraction framework based on interval neural networks to verify the ∆-robustness of CFXs against a possibly infinite set of changes to the model parameters, i.e., weights and biases. We then demonstrate the utility of this approach in two distinct ways. First, we analyse the ∆-robustness of a number of CFX generation methods from the literature and show that they unanimously host significant deficiencies in this regard. Second, we demonstrate how embedding ∆-robustness within existing methods can provide CFXs which are provably robust.