RESUMEN

OBJECTIVE: To investigate how the general public trades off explainability versus accuracy of artificial intelligence (AI) systems and whether this differs between healthcare and non-healthcare scenarios. MATERIALS AND METHODS: Citizens' juries are a form of deliberative democracy eliciting informed judgment from a representative sample of the general public around policy questions. We organized two 5-day citizens' juries in the UK with 18 jurors each. Jurors considered 3 AI systems with different levels of accuracy and explainability in 2 healthcare and 2 non-healthcare scenarios. Per scenario, jurors voted for their preferred system; votes were analyzed descriptively. Qualitative data on considerations behind their preferences included transcribed audio-recordings of plenary sessions, observational field notes, outputs from small group work and free-text comments accompanying jurors' votes; qualitative data were analyzed thematically by scenario, per and across AI systems. RESULTS: In healthcare scenarios, jurors favored accuracy over explainability, whereas in non-healthcare contexts they either valued explainability equally to, or more than, accuracy. Jurors' considerations in favor of accuracy regarded the impact of decisions on individuals and society, and the potential to increase efficiency of services. Reasons for emphasizing explainability included increased opportunities for individuals and society to learn and improve future prospects and enhanced ability for humans to identify and resolve system biases. CONCLUSION: Citizens may value explainability of AI systems in healthcare less than in non-healthcare domains and less than often assumed by professionals, especially when weighed against system accuracy. The public should therefore be actively consulted when developing policy on AI explainability.

Asunto(s)

RESUMEN

Terrestrial organisms that use traveling waves to locomote must leverage heterogeneities to overcome drag on the elongate body. While previous studies illuminated how habitat generalist snakes self-deform to use rigid obstacles in the surroundings, control strategies for multi-component terrain are largely unknown. We compared the sand-specialist Chionactis occipitalis to a habitat generalist, Pantherophis guttatus, navigating a model terrestrial terrain-rigid post arrays on a low-friction substrate. We found the waveshapes used by the generalist were more variable than the specialist. Principal component analysis revealed that while the specialized sand-swimming waveform was always present on C. occipitalis, the generalist did not have a similarly pervasive low-dimensional waveshape. We expected the generalist to thus outperform the specialist in the arrays, but body slip of both species was comparable on level ground and in all trials the snakes successfully traversed the arena. When we further challenged the snakes to ascend an inclined lattice, the sand-specialist had difficulty maintaining contact with the obstacles and was unable to progress up the steepest inclines in the largest lattice spacings. Our results suggest that species adapted to different habitats use different control modalities-the specialist is primarily controlling its kinematics to achieve a target shape while, consistent with previous research, the generalist is using force control and self-deforms in response to terrain contacts. While both strategies allowed progress on the uninclined low-friction terrain with posts, the more variable waveshapes of the generalist may be necessary when faced with more challenging locomotor tasks like climbing inclines.

RESUMEN

While terrestrial locomotors often contend with permanently deformable substrates like sand, soil, and mud, principles of motion on such materials are lacking. We study the desert-specialist shovel-nosed snake traversing a model sand and find body inertia is negligible despite rapid transit and speed dependent granular reaction forces. New surface resistive force theory (RFT) calculation reveals how wave shape in these snakes minimizes material memory effects and optimizes escape performance given physiological power limitations. RFT explains the morphology and waveform-dependent performance of a diversity of non-sand-specialist snakes but overestimates the capability of those snakes which suffer high lateral slipping of the body. Robophysical experiments recapitulate aspects of these failure-prone snakes and elucidate how re-encountering previously deformed material hinders performance. This study reveals how memory effects stymied the locomotion of a diversity of snakes in our previous studies (Marvi et al., 2014) and indicates avenues to improve all-terrain robots.

Asunto(s)

RESUMEN

Limbless animals like snakes inhabit most terrestrial environments, generating thrust to overcome drag on the elongate body via contacts with heterogeneities. The complex body postures of some snakes and the unknown physics of most terrestrial materials frustrates understanding of strategies for effective locomotion. As a result, little is known about how limbless animals contend with unplanned obstacle contacts. We studied a desert snake, Chionactis occipitalis, which uses a stereotyped head-to-tail traveling wave to move quickly on homogeneous sand. In laboratory experiments, we challenged snakes to move across a uniform substrate and through a regular array of force-sensitive posts. The snakes were reoriented by the array in a manner reminiscent of the matter-wave diffraction of subatomic particles. Force patterns indicated the animals did not change their self-deformation pattern to avoid or grab the posts. A model using open-loop control incorporating previously described snake muscle activation patterns and body-buckling dynamics reproduced the observed patterns, suggesting a similar control strategy may be used by the animals. Our results reveal how passive dynamics can benefit limbless locomotors by allowing robust transit in heterogeneous environments with minimal sensing.

Asunto(s)