Has China beat the West to detailed AI governance rules?

Oct 5, 2021 | Jim Nash

Has China beat the West to detailed AI governance rules?

A number of publications globally are reporting that China has approved its first AI industry ethics guidelines.

It is unclear if the guidelines — six principles — have the force of law. The government may have removed the document from public view, as it could not be found Monday on any site.

Multiple publishers have reported that China’s Ministry of Science and Technology published the New Generation Artificial Intelligence Ethics Specifications on September 27.

The guidelines are described in media reports as being among the more specific that have been enacted worldwide, but based on these descriptions, the rules still feel conceptual.

UK tech and business publication Verdict quoted a recommendation of the document as stating that artificial intelligence needs to be “people-oriented and follow the common values of mankind”; hardly the stuff of governance roadmaps.

The South China Morning Post, citing a Chinese document that its editors presumably have seen, reported that the guidelines put humans in control of AI algorithms, AI services and interactions with AIs.

Verdict reported that, along with keeping humans in the decision-making process, the document promotes “AI ethics and behavioural consciousness for the whole society.”

There is at least one other, seemingly more useful, ongoing AI ethics effort underway, this time in Europe. A private/public project has been launched in the Netherlands to see if an ethics tool can be used to mitigate risks involved in the deployment of facial recognition by law enforcement. The European Commission is also developing regulations around AI which could be used as a benchmark elsewhere, as has been the case for the bloc’s GDPR data protection regulation.

Given the Chinese government’s nationwide biometric surveillance network and its digital-identity Social Credit System, a black list that continuously scores citizens for their trustworthiness, the guidelines likely only apply to non-government systems use.

It is also safe to assume that any restraint imposed by Beijing on domestic AI and machine learning companies does not apply to systems aggressively sold by the same vendors overseas.