The former Texas congressman’s plan takes an expansive view of both the potential and the risks of artificial intelligence, calling for it to be used more widely but also tightly regulated.
The policy plan on artificial intelligence released by former Representative Will Hurd of Texas on Wednesday makes him the first candidate in the Republican presidential field to formally propose a way to navigate the uses and dangers of a technology so thorny he likened it to nuclear fission.
“Nuclear fission controlled gives you nuclear power — clean, cheap, limitless power,” Mr. Hurd said in an interview with The New York Times. “Nuclear fission uncontrolled gives you nuclear weapons that can destroy the world. And I think A.I. is equivalent.”
The plan, first reported by Axios, takes an expansive view of both the potential and the risks of artificial intelligence. He calls for A.I. to be used much more widely than it currently is — both in administrative tasks within the federal government and in highly sensitive areas like national defense — but also supports regulating the industry more tightly than is typical of many Republicans’ approach to private industries.
Among his proposals are calls to ensure compensation when people’s intellectual property is used in A.I.-generated content, as well as name, image and likeness protections against so-called deepfakes. He would also seek to require permits for companies that want to build A.I. models and to impose “strict regulations” on exports of A.I. technology, and would reject any exemptions for developers from liability under existing laws.
Artificial intelligence has already begun to change political campaigns themselves, with some operatives using it to write first drafts of fund-raising messages and automate tedious tasks — and to spread disinformation, including fake images of opponents.
Mr. Hurd has struggled to gain traction in the Republican primary field. He did not qualify for the first debate in August because he failed to reach 1 percent support in enough polls, and he remains at risk of failing to meet the even higher thresholds to qualify for the second debate next week.
But that he would be the first candidate to release a formal plan on artificial intelligence tracks with his professional background.
He once worked as a senior adviser at a cybersecurity firm called FusionX, and made cybersecurity one of his main focuses as a legislator. He also led the House Oversight Subcommittee on Information Technology, where he organized hearings on artificial intelligence in 2018, long before it entered the mainstream. After leaving Congress in 2021, he joined the board of OpenAI, the artificial intelligence laboratory that developed ChatGPT.
“Artificial intelligence is a technology that transcends borders,” Mr. Hurd said at the first congressional A.I. hearing in 2018. “We have allies and adversaries, both nation-states and individual hackers, who are pursuing artificial intelligence with all they have because dominance in artificial intelligence is a guaranteed leg up in the realm of geopolitics and economics.”
His plan suggests employing A.I. tools within military, intelligence and border security agencies and using those tools to make the government “more responsive to the needs of everyday Americans.” He said in the interview that this could include using A.I. to issue passports and visas, summarize publicly available information for intelligence agencies, predict what federal support individual communities will need as a hurricane approaches and identify the cause of backlogs at poorly performing Veterans Affairs centers.
Current A.I. models have a well-documented tendency to “hallucinate” and provide inaccurate or fabricated information. Mr. Hurd’s plan does not address that problem. He said he envisioned A.I. helping migrants learn English and helping students with math, and was “not as concerned” with hallucination in those contexts.
“I think we can achieve the promise of A.I. while minimizing the risk,” he said.
Source: Elections - nytimes.com
