The G7 on AI; not A1.

This is the G7 Innovation Minister’s statement on Artificial Intelligence from the G7 meeting in Charlevoix in May of this year. Once again I’m a little late to the party even though it happened in Canada.

Frankly it’s a trite and disappointing hodge-podge of low quality ideas with no central theme linked by the liberal use of semi-colons and the occasional tossed-in setence fragment. Here’s a particular gem: “Supporting economic growth from AI Innovation is about using AI applications to help improve economic performance.” You don’t say. And no doubt using AI applications to help improve economic performance is about supporting economic growth from AI Innovation. Some goal statements, like opposing data localization requirements that are unjustifiable, aren’t even really on all fours with reality, given that some G7 members already have legislation in place to deal with this in a much more direct fashion via the EU.

Untangling the rest of the preamble is not worth the effort as it uses an astonishing number of words to say very little.

On the legal and regulatory front, the big push is on advancing “appropriate technical, ethical and technologically neutral approaches by: safeguarding privacy; investing in cybersecurity, the enforcement of applicable privacy legislation and communication of enforcement decisions; informing individuals about existing bodies of national law, including in relation to how their personal data may be used by AI systems; promoting R&D by industry in safety, assurance, data quality, and data security; and exploring the use of other transformative technologies to protect personal privacy and transparency.”

Not much about this seems to be on point on anything. The extreme focus on privacy seems over the top, though I suppose that is what many people in the public are worried about and given that the statement never meaningfully defines what AI is, I guess it could encapsulate any kind of software. So sure, go whole hog on privacy.  I have no idea whether “neutral approaches” qualifies both of “ethical and technologically” and given the rest of the puncuation sprinkled about one wonders if someone needed to borrow an oxford comma there. Assuming it is just “technologically neutral” it seems like an out of place goal that really wouldn’t be driven by the subsequent focus on privacy and transparency. Really technological neutrality would be a question of establishing appropriate regulatory processes. Maybe that is an assumed elephant in the room. And the goal in itself seems to have wandered in from the cold given that nothing in the preamble really sets it up as important. Whether the bodies of national law that are referred to are bodies of privacy law or bodies of law dealing with AI is a question unanswered.

The statement proposes that it be disseminated globally, which given the content, I would discourage.