(Bloomberg) --US lawmakers on Thursday questioned how new rules for artificial intelligence can protect against the technology's risks without reinforcing the early advantages enjoyed by tech giants such as Microsoft and Google.
Representative Frank Lucas, the Oklahoma Republican who chairs the House Science, Space and Technology Committee, opened Thursday's hearing by recognizing the need for Congress to set some guardrails for AI while warning that any regulation must also promote innovation — especially as the US races China to develop machine learning capabilities.
Shahin Farshchi, a general partner of venture capital firm Lux Capital, warned the panel that a few companies are already dominating a sector in which it can cost more than $100 million to train the most advanced generative AI models. Microsoft Corp., Alpahabet Inc.'s Google and Amazon.com Inc. are also among the few companies that have the computing power to run the most powerful models.
"There are steep barriers to entry for AI researchers and founders," Farshchi said. "If we do not provide open, fair and diverse access to computing resources, we could see a concentration of resources in a time of rapid change, reminiscent of Standard Oil during the industrial revolution."
Congress is rushing to understand just what AI is and how to protect against potential risks such as bias, threats to democracy and misuse by nefarious actors. The conversation is taking place as global adversaries like China develop their own AI tools that have the potential to transform economies and shape future conflicts.
Some lawmakers have warned that the costs of complying with government regulations would hinder smaller startups' ability to compete with tech giants.
Senate Majority Leader Chuck Schumer announced plans Wednesday for a series of forums on AI and set a timeline to draft legislation to regulate the technology within "months, not years."
Ted Lieu, a California Democrat who has introduced his own AI-focused legislation, asked whether open source technology would foster US innovation, only to have those advances benefit global adversaries.
Clement Delangue, co-founder and CEO of Hugging Face, an AI startup, told the House panel that openness is essential to preserve broad and vibrant competition. That in turn will better position the US to innovate in the race against China, he added.
"Open science and open source distribute economic gains by fostering tens of thousands of startups building with AI, in all domains and sectors" Delangue said. "It fosters innovation, as we have seen in the past few years, and fair competition between all."
Some companies, like Microsoft, have suggested a licensing regime as a way for the government to regulate the most powerful AI models. Democratic Representative Zoe Lofgren, whose California district includes part of Silicon Valley, suggested requiring registration of AI models as a way to provide accountability without advantaging the companies that would be best positioned to win licenses.
