There is general consensus that it is important for artificial intelligence (AI) and machine learning systems to be explainable and/or interpretable. However, there is no general consensus over what is meant by 'explainable' and 'interpretable'. In this paper, we argue that this lack of consensus is due to there being several distinct stakeholder communities. We note that, while the concerns of the individual communities are broadly compatible, they are not identical, which gives rise to different intents and requirements for explainability/interpretability. We use the software engineering distinction between validation and verification, and the epistemological distinctions between knowns/unknowns, to tease apart the concerns of the stakeholder communities and highlight the areas where their foci overlap or diverge. It is not the purpose of the authors of this paper to 'take sides' -we count ourselves as members, to varying degrees, of multiple communities -but rather to help disambiguate what stakeholders mean when they ask 'Why?' of an AI.

No abstract

No abstract

Understanding the effect of horseshoe-surface combinations on hoof kinematics at gallop is relevant for optimising performance and minimising injury in racehorse-jockey dyads. This intervention study assessed hoof breakover duration in Thoroughbred ex-racehorses from the British Racing School galloping on turf and artificial tracks in four shoeing conditions: barefoot, aluminium-rubber composite (GluShu), aluminium and steel. Shoe-surface combinations were tested in a randomized order and horse-rider pairings (n=14) remained constant. High-speed video cameras (Sony DSC-RX100M5) filmed the hoof-ground interactions at 1000 frames per second. The time taken for a hoof marker wand fixed to the lateral hoof wall to rotate through an angle of 90 degrees during 384 breakover events was quantified using Tracker software. Data were collected for leading and non-leading front and hind limbs, at gallop speeds ranging from 23–56 km h-1. Linear mixed-models assessed whether speed, surface, shoeing condition or any interaction between these parameters (fixed factors) significantly affected breakover duration. Day and horse-rider pair were included as random factors and speed was included as a covariate. The significance threshold was set at p<0.05. For all limbs, breakover times decreased as gallop speed increased (p<0.0005), although a greater relative reduction in breakover duration for hindlimbs was apparent beyond approximately 45 km h-1. Breakover duration was longer on turf compared to the artificial surface (p≤0.04). In the non-leading hindlimb only, breakover duration was affected by shoeing condition (p=0.025) and an interaction between shoeing condition and speed (p=0.023). Future work seeks to relate these results to hoof accelerometer data.

Understanding the effect of horseshoe–surface combinations on hoof kinematics at gallop is relevant for optimising performance and minimising injury in racehorse–jockey dyads. This intervention study assessed hoof breakover duration in Thoroughbred ex-racehorses from the British Racing School galloping on turf and artificial tracks in four shoeing conditions: aluminium, barefoot, aluminium–rubber composite (GluShu) and steel. Shoe–surface combinations were tested in a randomized order and horse–jockey pairings (n = 14) remained constant. High-speed video cameras (Sony DSC-RX100M5) filmed the hoof-ground interactions at 1000 frames per second. The time taken for a hoof marker wand fixed to the lateral hoof wall to rotate through an angle of 90 degrees during 384 breakover events was quantified using Tracker software. Data were collected for leading and non-leading forelimbs and hindlimbs, at gallop speeds ranging from 23–56 km h−1. Linear mixed-models assessed whether speed, surface, shoeing condition and any interaction between these parameters (fixed factors) significantly affected breakover duration. Day and horse–jockey pair were included as random factors and speed was included as a covariate. The significance threshold was set at p < 0.05. For all limbs, breakover times decreased as gallop speed increased (p < 0.0005), although a greater relative reduction in breakover duration for hindlimbs was apparent beyond approximately 45 km h−1. Breakover duration was longer on turf compared to the artificial surface (p ≤ 0.04). In the non-leading hindlimb only, breakover duration was affected by shoeing condition (p = 0.025) and an interaction between shoeing condition and speed (p = 0.023). This work demonstrates that speed, ground surface and shoeing condition are important factors influencing the galloping gait of the Thoroughbred racehorse.