putilov_denis - stock.adobe.com

Experts see positives to China's new AI regulations

In the international push to regulate AI, China may have developed some of the most ambitious rules to date, according to Alan Pelz-Sharpe, founder of Deep Analysis.

China's new AI regulations tackle some of the most controversial elements of the technology, including algorithmic explainability and transparency -- something businesses will need to pay attention to, analysts said.

The AI regulations took effect March 1. They require AI algorithms that make recommendations to users to be moral, accountable and transparent, according to Guannan Lu, analyst at Forrester Research.

China's AI regulations prohibit algorithms from using personal data to change a product's price from one person to the next. Businesses will also be required to be transparent about the purpose of the algorithm they're using, such as for making product recommendations, according to a translation of China's AI regulations published by Stanford University.

China's new AI regulations are some of the most ambitious AI regulations to date, said Alan Pelz-Sharpe, founder of research firm Deep Analysis.

"There are elements in here that other nations could learn from," Pelz-Sharpe said.

Learning from China's approach 

Though it may be "counterintuitive" to give China credit regarding its AI regulations given the political climate, Pelz-Sharpe said its rules "set a new benchmark that may drive improved regulations elsewhere."

The AI regulations outlaw algorithmic price gouging and require an explainable algorithmic decision-making process. The algorithms have to be "trustworthy AI," meaning built and tested to be as fair, explainable and bias-free as possible, Pelz-Sharpe said.

While there are discussions about trustworthy AI in areas such as Europe and the U.S., Pelz-Sharpe said there's little from a regulatory standpoint when it comes to defining the phrase. China, he said, is "making this concrete."

"Beyond blatant and deliberate misuse of AI, two of the biggest factors of concern for its use are bias and explainability," he said. "These regulations at least attempt to address some of these concerns."

The AI sector, governments and regulators need to cut through the political fog and look closely at these new regulations.
Alan Pelz-SharpeFounder, Deep Analysis

Pelz-Sharpe said time will tell how well the AI regulations work and how enforceable they will be. But for now, there are good portions of the AI regulations for other countries drafting their own AI regulations to consider. Though the European Union has proposed draft AI regulations, the U.S. has not.

China's regulations provide a "blueprint for AI regulations elsewhere," Pelz-Sharpe said. Forrester's Lu  said China provides one example of the kinds of regulations that might be needed around the technology.

"Countries and regional authorities are learning from each other and promote their version iteratively," Lu said. "Others may learn from how China has set up clauses and its regulation enforcement."

Impact to businesses

Now that the regulations are in effect, businesses with operations in China that collect user data will need to closely assess how that data is used, Lu said.

The AI regulations will bring changes to business operations for companies in industries ranging from social media to e-commerce, which means business owners will need to not only assess their business models, but pricing strategies as well, Lu said.

"Firms should be more transparent on the intention of the algorithm and how their algorithms are using the data," he said. "This may change their consent terms and require new ways to inform customers. For some circumstances, firms may collect less data than before. That means they need to redesign the algorithms to run with less featured data."

Makenzie Holland is a news writer covering big tech and federal regulation. Prior to joining TechTarget, she was a general reporter for the Wilmington StarNews and a crime and education reporter at the Wabash Plain Dealer.

Dig Deeper on AI technologies

SearchBusinessAnalytics
SearchCIO
SearchDataManagement
SearchERP
Close