This principle of human freedom and autonomy connects with the principles of promoting the common good and preventing harm, and underscores all the more the importance of the principle of human-centering, freedom, and autonomy. [...] The purpose of a learning system and the framing of the problem it addresses, for example, should influence the choice of model used (whether or not to allow the choice of a more opaque model in terms of the ways it makes decisions). [...] We propose a risk-management model that requires a double observation in order to assess the dangerousness of a system, first to assess the potential dangerousness of the system as designed and then to assess the strength of the alignment of its task with its outcome, namely, the potential of a system to manifest its dangerousness outside the role its designers intended for it. [...] At the present time—at least for the next few years, until the field stabilizes—the AI regulator should be established in the form of a unit within the Regulatory Authority, because the latter’s roles are highly suited to the evolving world of AI regulation. [...] Right now, even in the absence of a broad and dedicated artificial intelligence law, designated decision-makers and regulators should be obligated to update existing legislation and pass supplementary legislation—mainly to statutes such as the Competition and Consumer Protection Law, the Copyright Law, the Protection of Privacy Law, the Evidence Ordinance, and the Government Procurement Law.
- Pages
- 15
- Published in
- Israel
