US Senate panel wants to get tough on AI, but new rules could risk America’s tech edge

WASHINGTON – It is not often that Singapore figures at a United States congressional hearing. 

The most notable instance in recent times was when TikTok’s Singaporean chief executive Chew Shou Zi testified before the House of Representatives in March to address concerns that China had pervasive influence over the platform.

On Tuesday, Singapore was mentioned at a Senate Judiciary Sub-committee hearing that met to consider how to regulate the game-changing field of artificial intelligence (AI).

A newly unveiled bipartisan proposal to regulate AI encourages Congress to utilise export controls, sanctions and other legal means to limit the transfer of high-performance chips, chip-making equipment and other advanced AI technologies to China, Russia, and other nations that the US regards as its adversaries.

But if the US over-regulates the sector, it might lose its edge to other nations, such as Singapore, the Senate heard from an expert testifying at the hearing. 

“There’s a careful balance to be made between limiting where your chips go and what they’re used for,” said Mr William Dally, the chief scientist and senior vice-president of research at Nvidia, a leading producer of AI chips and software.

Disadvantaging American companies, and the whole food chain that feeds them, will hurt, he said.

“If people can’t get the chips they need to do AI from us, they will get them somewhere else,” said Mr Dally.

“If all of a sudden, the standard chips for people who do AI become (those) from... pick a country, Singapore... all the software engineers will start writing all the software for those chips. They’ll become the dominant chips. 

“And the leadership of that technology area will have shifted from the US to Singapore,” he said.

Singapore has been in the headlines this week: US-based GlobalFoundries inaugurated its new US$4 billion (S$5.4 billion) microchip fabrication plant in Woodlands. It adds to Singapore’s total semiconductor output, which makes up 11 per cent of the global market.

Next to the Republican proposal on Tuesday to impeach President Joe Biden, regulating AI is now the hottest topic on Capitol Hill.

A series of closed-door meetings, convened by Senate Majority Leader Chuck Schumer, kicked off on Wednesday, featuring top figures from the world of AI – Tesla’s Elon Musk, Microsoft’s Bill Gates, Google’s Sundar Pichai, ChatGPT’s Sam Altman and others.

Described as brainstorming sessions, these will be attended by all 100 senators and will run through the current congressional session. 

At Tuesday’s hearing, Mr Dally said the participation of every major power was essential to create effective AI regulations.

“We should recognise that the next great AI applications may come from anywhere in the world,” he said.

No nation or company controls an AI “chokepoint”, he added. 

“Leading US computing platforms are competing with companies from around the world. While US companies may currently be the most energy-efficient, cost-efficient or easiest to use, they are not the only viable alternatives for developers abroad. 

“Other nations are developing AI systems, with or without US components, and they will offer those applications in the worldwide market. To ensure safe and trustworthy AI, we must engage with governments and stakeholders around the world.” 

Senators said there is a deep hunger for safeguards that encourage trust in AI. “There will be regulation; the question is how soon and what,” said Senator Richard Blumenthal, a Democratic lawmaker from Connecticut and chair of the sub-committee. 

While revolutionising industries and accelerating economic growth, it is feared that AI may also eliminate millions of jobs, become a conduit for interference in elections, and pose national security threats.

Mr Blumenthal and his Republican counterpart, Senator Josh Hawley from Missouri, last week released a bipartisan framework that recommends that AI companies be held legally liable over breaches of privacy.

Congress must also take steps to directly prohibit harms such as deepfake images. 

In enacting rules governing AI, the US lags behind many nations. The European Union is in the final stages of passing a comprehensive AI law, while China has issued sweeping oversight measures, including rules for generative AI, which came into effect in August. Asean is also in the midst of drawing up an AI guide for release in early 2024. 

Several top AI firms, including Nvidia and Amazon, have embraced a voluntary AI code of conduct espoused by the White House earlier in 2023.

Traditional command-and-control models of regulation, such as hard and prescriptive regulation, are not going to be effective in the tech marketplace, said Mr John Costello, deputy director at the Wadhwani Centre for AI and Advanced Technologies at the Washington-based Centre for Strategic and International Studies.

A co-regulatory approach, where the sector has some responsibility for setting its own standards that are enforced by a regulatory agency, might be better.

Such models can keep up with the pace of change in technology, he said.