Niveau de dépendance des enfants et adolescents porteurs d'autisme. Suivi longitudinal à cinq ans

An interdisciplinary conceptual study of Artificial Intelligence (AI) for helping benefit-risk assessment practices

“…Applied to AI devices, the dependence level should be envisaged first as regard to human intervention on machine functioning and decision-making, then, as regard to the natural/technological environment in which the device operates. Psychologists involved in autism [10] identified three levels of dependence based on the assessment of defined behavioural items characterising human autonomy (known as ESDA 3 , [9]) which could inspire AI stakeholders: 1-partial and discontinuous dependence; 2-partial and continuous dependence; 3-full dependence. This scale could be used regarding general functioning of a device or regarding specific characteristics such as cognitive abilities' implementation.…”

“…It supposes an information processing module intended to make the data understandable for the machine. In any case, AI devices must be legally qualified as data processing 10 devices. Hence existing data protection laws applying to personal and eventually public data [44] must be respected.…”

“…Section 1 regarding the concept of "intelligence". 10 A data processing can be understood by reference to the definition provided by the EU General Data Protection Regulation 2016/679 (GDPR) Article 4(2) which, by extrapolation, can usefully apply to all data. The GDPR defines data processing as "any operation or set of operations which is performed on […] data or on sets of […] data, whether or not by automated means, such as collection, recording, organisation, structuring, storage, adaptation or alteration, retrieval, consultation, use, disclosure by transmission, dissemination or otherwise making available, alignment or combination, restriction, erasure or destruction".…”

“…But an AI device can help humans to convey a human-sourced legal act, to formalise or to implement legal acts by being driven by programmed procedures and criteria, in an automated way (e.g. with smart contracts [10]). Therefore, considering a machine as personally accountable is a fiction, the machines being anyway programmed by a human they should only be tools under the human responsibility whose calculated solutions, even if self-determined within a panel of options, are used as a support of human purposes, in human decision-making or into human-controlled environments, under humans' laws such as contract law.…”