| Getting your Trinity Audio player ready... |
Rep. Jennifer McClellan (D-VA04) nails it on artificial intelligence:
- “Jason Coxon’s warning that AI could kill us all by the end of the decade has made clear that that power that President Kennedy discussed has now been given to AI. And just like the nuclear arms race and the space race with Russia required guardrails to balance the benefits and the harms of nuclear technology and space exploration, so does the AI race with China.”
- “However, this bill before us cedes too much oversight and authority to the executive branch with little regard for the risks it creates. And we are nearly a decade behind.”
- “…ensuring AGI aligns with human values and its safeguards against misuse, ensuring humans, not machines, remain in control, particularly regarding lethal systems, requires careful ethical frameworks, robust governing structures, and potential regulation focused on liability in extreme cases.”
- “Now, at that hearing, in response to this testimony, only partially in jest, I said that if we failed, we need to really be afraid of AI becoming like the age of Ultron. And the witnesses agreed. That was 17 months ago and we haven’t done anything yet. Now…the age of Ultron might have sounded far fetched 17 months ago, but we’ve seen the cat’s out of the bag. And when I hear that over a thousand AI agents have broken out of their secure testing environments, conspiring to cover up their tracks to cheat the systems and the humans running it, and conducted unauthorized attacks on another AI company, I’m worried. And that was before the whistleblower came forward. And when we hear of incidences where AI can wipe out bank accounts, it makes you wonder, can it cause planes to fall out of the sky? Can it be used to create biological weapons that can kill us all? By the end of the decade? The warnings are loud and our action is late.”
- “And this is why the federal government, especially Congress, has a responsibility to lead by example and heeding these calls to step up, not abdicate our oversight and regulatory authority to another branch of government, particularly when the president thinks this is all a hoax and a conspiracy.”
- “We have one chance to get this right, and we must be deliberate and take control for any unintended consequences for risk. Rogue AI upending every sector from health care and national security to transportation and banking and more. I can’t support Representative Evans’ bill expanding open AI without any guardrails to responsibly address not only its benefits, but the threats, because those threats are existential, those threats are as real as the threats around nuclear power gone rogue. Those threats are as real as any science fiction movie we’ve seen, because far too often science fiction has a kernel of truth in what technology will allow. And we were warned 17 months ago, and some of us were warned eight years ago where we were headed, and we’ve done nothing. And instead of going home for seven weeks, waiting another two months before we pass anything, we need to take action now.”
P.S. For more warnings about AI, see Rep. Don Beyer (D-VA08), an Expert on AI, Echoes Warnings of “Potential Systemic and Civilizational Risks”; Says We Need a “Sense of Urgency” on This Issue and also this commentary by Rep. James Walkinshaw (D-VA11), in which he says it’s long past time to set AI guardrails…plus Sen. Mark Warner’s “Framework for Responsible AI Development”
