Submission 104
Dr Ben Saunders
Submitter group: Individuals.
Position labels are not shown for individual submitters.
An AI model identified the submitter group and wrote the summary, key points and recommendation summaries. Counts do not measure importance or merit.
Summary
Read the original submission for the full text.
Proposes statutory provisions, which it says would need drafting refinement, making the creator of an AI technology civilly and criminally responsible for actions the technology performs that would be actionable or unlawful if done by a person. Also argues trial-and-error targeted regulation by executive order, with defined and reviewable delegations, may suit AI better than traditional law-making.
Key points
Quotes are the submitter’s own words. Text taken from documents, including scanned pages read by text-recognition software, can differ from the PDF. Check the original. Listed in the same order as on the submission’s page in the interactive guide.
Proposes a civil provision treating the creator of an AI technology as having committed actions that would give rise to a civil cause of action if done by a person.
Where an action is performed by an artificial intelligence technology, and that action would give rise to a civil cause of action if performed by a natural person, the person responsible for creating the artificial intelligence technology is taken to have committed the action.
Proposes a parallel criminal provision making the AI creator responsible for conduct that would be an offence if performed by a natural person.
Where an action is performed by an artificial intelligence technology, and that action would constitute unlawful conduct or an offence if performed by a natural person, the person responsible for creating the artificial intelligence technology is taken to have committed an offence.
Argues that the proposed provisions would establish creator accountability, provided definitional and other drafting issues could be satisfactorily resolved.
Provided those definitional and other issues could be satisfactorily resolved, these proposed statutory provisions would force the producers of AI technologies to take responsibility for their creations and provide clear legal accountability for the actions of AI technology.
Argues that rapidly implemented targeted regulations, amendable or withdrawable if ineffective, may suit AI better than traditional law-making.
Rather, a process of trial and error regulating may be more appropriate, whereby targeted regulations are rapidly implemented to deal with specific problems relating to AI; those regulations could be quickly amended or withdrawn if they prove to be ineffective.
Says broad delegations of executive power are likely to raise concerns about accountability and liberty, and should be carefully defined in scope and subject to review and oversight.
Broad delegations of executive power are (rightly) likely to raise concerns about accountability and liberty. Any delegations should be carefully defined in scope (especially as to time) and subjected to mechanisms of review and oversight.
Recommendations
Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.
Enact a provision along the proposed lines deeming AI creators responsible for actions that would give rise to a civil cause of action if performed by a natural person.
Where an action is performed by an artificial intelligence technology, and that action would give rise to a civil cause of action if performed by a natural person, the person responsible for creating the artificial intelligence technology is taken to have committed the action.
Enact a provision along the proposed lines deeming AI creators to have committed an offence when AI actions would constitute unlawful conduct or an offence if performed by a natural person.
Where an action is performed by an artificial intelligence technology, and that action would constitute unlawful conduct or an offence if performed by a natural person, the person responsible for creating the artificial intelligence technology is taken to have committed an offence.
Resolve definitional and other drafting issues before making the proposed liability provisions workable law.
Evidently, some drafting finesse would be required to render these into workable statutory provisions. Provided those definitional and other issues could be satisfactorily resolved, these proposed statutory provisions would force the producers of AI technologies to take responsibility for their creations and provide clear legal accountability for the actions of AI technology.
Consider broad delegations of power to the executive, with targeted regulation by executive order, as a way to deal with AI regulation.
Such targeted regulations would be more appropriately implemented by executive orders than primary legislation. I therefore propose that broad delegations of power to the executive may be a preferable way to deal with the problem of AI regulation.
Define any executive delegations carefully in scope, especially time, and subject them to review and oversight.
Any delegations should be carefully defined in scope (especially as to time) and subjected to mechanisms of review and oversight.