China Standardization

京公网安备 11011402011482号

京ICP备12047668号-3
Hot Spots
Meetings
ANSI World Standard Week 2021
Recommendation
Your Position:Home > News >
Standardization contributes to a framework for AI
IEC 2021-10-22

Artificial intelligence has a lot in common with dogs, at least when it comes to the law. The canine analogy came during an IEC Workshop, co-organized with the Directorate of International Law of Switzerland's Federal Department of Foreign Affairs.

"My dog used to be a guide dog but it's not anymore, so I can't take it on a train or a plane or into a café," said Philippa Ryan, Associate Professor, Australian National University College of Law. She was addressing the IEC Workshop on Artificial Intelligence and the Interplay between Standardization, Conformity Assessment and Legislation.

"If I cross the border from New South Wales into Victoria my dog needs to wear a seatbelt, so the rules change depends on where he is, and it depends on his job. Meanwhile, it also depends on what I trained him to do: if I train him to kill small rabbits then most people in the country won't care, but if I trained him to kill my next-door neighbor's pet bunny, then it's going to be a problem".

Professor Ryan explained that much like her dog, the regulatory environment in which artificial intelligence operates came down to context and the relationship with humans.  

In his presentation, Ambassador Roger Dubach of the Swiss Federal Department of Foreign Affairs identified four reasons why AI is attracting attention at the level of international law. "It's very relevant to discuss AI because behind all technologies are values and the export of technology equals the export of values," argued Ambassador Dubach.

"Secondly, it's hugely strategic and as AI is seen as strategic it heavily influences the legal discussions or in general, the regulatory discussions. The third point is that the relationship between human beings and machines is now changing.

"With AI systems, traditionally law was made by humans for humans to be applied to humans. Now we have machines that are able to make recommendations, which gives rise to some fundamental legal questions.

"And the fourth point is that AI systems enable completely new applications such as facial recognition and so there's also a huge discussion about these new applications".

The Vice Director of the Swiss Federal Institute of Metrology, Bobjoseph Mathew, emphasised the importance of conformity assessment. "It is really important to build trust, otherwise you cannot reap the benefits of AI," he said.

"It is important to maximize the benefits of AI, but this is only possible if we have a way to overcome the risks or to mitigate the risks and the compliance issues. One instrument for this is conformity assessment that demonstrates that a product or system meets specific requirements in the standards that are used."

Wael Diab presented an overview of the development of standards for AI in the joint technical committee set up by IEC and ISO. Mr Diab, who chairs ISO/IEC JTC 1/SC 42, highlighted the interdisciplinary nature of the standardization work.

Swedish lawyer Viveka Bonde, a member of SC 42, cited the example of the contributors to a new document providing an overview of ethical and societal concerns. "We have participation from, of course, the tech people, but we also have ethicists, and we have even business angels, we have lawyers, we have financial advisors, so we have a wide spectrum," she said.

Google's Lenora Zimmerman, another expert working in SC 42, explained that when developing a standard, it was important to look at the work from the perspective of stakeholders. "You need to identify trustworthiness concerns to address at each stage and for each one of the stakeholders," she said. "We're thinking about it at the higher level of who it is going to impact and what it is going to be used for and what are the considerations that need to be taken into account".

Closing the workshop, IEC General Secretary Philippe Metzger said the dialogue initiated by the workshop would continue at an AI conference jointly organized by the IEC and the Swiss Federal Department of Foreign Affairs, in Geneva, on 16 and 17 May 2022. He said the work being carried out by SC 42 and elsewhere in the IEC could support and complement the merging international framework for AI.

(Source: IEC)

Home Page