Congress To Consider Two New Bills On AI

An anonymous reader quotes a report from Reuters: U.S. senators on Thursday introduced two separate bipartisan artificial intelligence bills on Thursday amid growing interest in addressing issues surrounding the technology. One would require the U.S. government to be transparent when using AI to interact with people and another would establish an office to determine if the United States is remaining competitive in the latest technologies. Senators Gary Peters, a Democrat who chairs the Homeland Security committee, introduced a bill along with Senators Mike Braun and James Lankford, both Republicans, which would require U.S. government agencies to tell people when the agency is using AI to interact with them. The bill also requires agencies to create a way for people to appeal any decisions made by AI.

"The federal government needs to be proactive and transparent with AI utilization and ensure that decisions aren't being made without humans in the driver's seat," said Braun in a statement. Senators Michael Bennet and Mark Warner, both Democrats, introduced a measure along with Republican Senator Todd Young that would establish an Office of Global Competition Analysis that would seek to ensure that the United States stayed in the front of the pack in developing artificial intelligence. "We cannot afford to lose our competitive edge in strategic technologies like semiconductors, quantum computing, and artificial intelligence to competitors like China," Bennet said.

Earlier this week, Senate Majority Leader Chuck Schumer said he had scheduled three briefings for senators on artificial intelligence, including the first classified briefing on the topic so lawmakers can be educated on the issue. The briefings include a general overview on AI, examining how to achieve American leadership on AI and a classified session on defense and intelligence issues and implications. Further reading: Ask Slashdot: What Are Some Good AI Regulations?

Too early.

[backslashdot]

Not only is this bill premature (like Congress in the horse & buggy era trying to come up with automobile regulations), it's also very dangerous and wrong.z

Re:

[TwistedGreen]

I agree.z

Re:

[account_deleted]

Comment removed based on user account deletion

Re:

[Thoth Ptolemy]

Why is it dangerous and wrong?

I have watched this movie before

[rtowne72]

Minority Report....or something like that.

Congress isn't going to pass anything

[rsilvergun]

At least not for the next year and a half. It's completely paralyzed by a handful of lunatics in the House of Representatives. The speaker of the House barely won his vote and to secure the votes he needed he had to agree to allow himself to be removed by anyone member of his party. So those lunatics have a ton of power and they're flexing it to show off to their extreme base in their gerrymandered districts. So they're doing things like shooting down every piece of legislation.

The Republican party is fracturing. They've even got a name for the factions, they called him the five families (no idea why that name was chosen). The Democratic party was always a coalition party so they've always had a hard time with it but the Republican Party was a basic alliance between the Evangelical extremists and The Business people, at least since Goldwater. Ordinarily the money people would step in and they would just primary all the crazies but the districts are so gerrymandered and Republican primary voter turnout is so high they haven't been able to do that. That's why we got crazy people like Herschel Walker and Marjorie Taylor green

Re:

[Powercntrl]

Ordinarily the money people would step in and they would just primary all the crazies

Did you miss Musk's tweet prior to the mid-terms, where he basically straight up admitted that a paralyzed government was precisely his desired outcome? These "crazies" didn't fund their campaigns on hopes and prayers, somebody obviously liked what they were selling. Never discount the fact that there probably are rich people happy to see a big ol' money wrench thrown into the gears of congress.

It was a $5 wrench

[Powercntrl]

That should've been monkey wrench, but I guess it also works as a Freudian slip.

Musk didn't want a paralyzed gov't

[rsilvergun]

that's just him showing off to the small gov't right wingers. It's part of a thing called "Starve The Beast". Musk's a billionaire. He wants the gov't to do stuff, but only for him. Not anyone else.

They didn't want those crazies. They cost them the mid terms. With the Senate and 20 seats in the house the GOP could've got cuts to social security, more food stamp cuts (the ones they got expire in 2 years and are a net *increase*) and of course another round of tax cuts for the rich. Meanwhile Musk could g

AI Warning Labels

[classiclantern]

Any legislation concerning ML or AI must be mind-numbing simple to be at all enforceable. My suggestion to lawmakers is; all art, writing, diagnosis, advice, hardware, software, music, plane schedules, video, photographs, or whatever the product is; tagged as being wholly or partly a product of computer manipulation. This lets the user know there is a high probably of error, the output is not fact-checked, and lets other AI know the source should not be relied-on as an MML input. One of the unsolved problems with AI is the information the AI uses to learn, was generated by another AI of unknown reliability. I predict the multiplication of errors in AI will eventually destroy the world as each output is a race to the bottom. Any and all programs that process data without human intervention must include a tag generated by the process itself, in the output. The public deserves to know what is AI and what is RI. There will always be a market for products marked as Real Intelligence and a standard symbol should be created for both cases. I realize this law will lead to a Prop 65 situation where everything will be marked as containing potentially dangerous AI but, the customer will usually have the option to choose the alternative 100% RI products at a higher cost.

Re:

[account_deleted]

Comment removed based on user account deletion

How would you draw the line?

[LeDopore]

If a live human customer service agent in a chat is mostly just rubber-stamping the advice their AI assistant gives them, does that qualify as a human interaction? There's no crisp line between content assisted by AI and generated by AI. Without such a crisp line it's hard to make an effective law.