Congress Told AI Holds Great Risks and Benefits for US Military
WASHINGTON — Artificial intelligence experts warned Tuesday during a congressional hearing of ominous dangers for the United States if it falls behind in developing the technology but a bright future by taking the lead.
One of the greatest risks would be defending against a foreign enemy whose robots and computers could overwhelm the U.S. military, they said.
“Artificial intelligence is going to be similar to nuclear weaponry,” said Alexandr Wang, chief executive officer of Scale AI, a San Francisco, California-based company that provides training data for artificial intelligence applications such as self-driving cars, mapping and robotics.
Unlike closely guarded nuclear weapons, “This technology is likely to be accessible to nearly everyone in the world,” he told the House Armed Services Subcommittee on Cyber, Information Technologies and Innovation.
AI refers to the capacity to program computers or robots to perform operations with the same learning and decision-making ability as humans, such as speech recognition or answering questions.
It holds great promise for some fields, such as helping paraplegics to walk, brain injury victims to compensate for their limited abilities and manufacturing plants to increase productivity through more sophisticated automation.
“We’re now embarking on a new era of the world,” Wang said.
For the military, it means machines could be used to decide which sites get destroyed or people get killed. It also could be used to disarm the automated weapons systems of adversaries.
Other countries are hastily pursuing development of the technology, Wang and other witnesses said.
China has set a goal of being the world leader in AI by 2030. Much of its effort would benefit the country’s military.
“Today I would say it’s a jump ball” on whether the United States or China has the best data needed to program AI systems, Wang said.
Klon Kitchen, a senior fellow at the American Enterprise Institute public policy foundation, said AI could usher in prosperity by adding $2.6 trillion to the global economy but that “it also has serious national security risks.”
Among the risks is the greater ease of cyberattacks by nearly anyone who gains access to it.
“Generative AI will soon pour gas on these fires,” Kitchen said.
Much of AI is being developed by private industry, where government controls are limited, he said.
“It applies, unfortunately, to the bad guys as well as the good guys,” Kitchen said.
Lawmakers at the hearing acknowledged their occasional confusion on how to manage the new technology.
Rep. Elissa Slotkin, D-Mich., described the challenge as determining “what are the ground rules for this new technology because it sounds scary.”
A recent Stanford University report showed that the annual number of AI-related laws passed in the 127 countries surveyed jumped from one in 2016 to 37 in 2022.
So far, federal legislative proposals are directed at portions of the technology. One bill would require AI-generated videos to include watermarks so viewers would know they are fakes. Another pending bill would require political ads to disclose they feature AI-generated content.
The Senate held its first briefing for top lawmakers on artificial intelligence last month in what they described as a first step toward a lengthy process of regulation.
Key Senate leaders, such as Senate Majority Leader Chuck Schumer, D-N.Y., said they are at least months away from any major legislative proposals to manage the benefits and risks of AI.
Inspiration for figuring out the military applications won a boost recently in Ukraine by “the buffoonery in the Russian military,” Slotkin said.
Haniyeh Mahmoudian, an ethicist for the machine learning software company DataRobot, Inc., said improperly programmed AI could turn against the people who are supposed to control it.
“We also need to invest in the guardrails,” Mahmoudian said. “It’s all about how we want to have our values embedded into these AI systems.”
You can reach us at [email protected] and follow us on Facebook and Twitter