This is so short sighted... autonomous vehicles including buses and trucks are on their way to our streets. We don't want to create rules and govern how this is going to work on our public roads? It's just going to be everyone for themselves, the vehicles will just follow rules meant for humans?<p>We have an opportunity here to set rules that cars should yield to rapid transit public buses, that vehicles should behave in ways to increase the flow of traffic, etc etc... there are many options for setting rules that autonomous vehicles must follow which is in the best interests of the public not just the rider.
> automated decision systems<p>So if a bank has an automated loan approval system that consists of a series of IF-THEN statements, and one of those statements amounts to IF (applicant.race != "White"), loan.reject; this ban would forbid a state from taking action?
Is this constitutional? It sounds like a pretty clear breach of the anti-commandeering doctrine. The federal government can't simply issue commands to state legislatures.<p>Federal law might supersede state law in areas where the federal government has express powers, e.g. interstate commerce, but if a state is adding AI-related provisions to existing policy in an area it already has authority over, I can't imagine how Congress could attempt to suppress that.<p>Sure, federal law could likely supersede state law if a state is trying to restrict AI as a commercial service in itself, as that would cross into interstate commerce territory. But if a state already has regulatory authority over e.g. how insurance companies operate within their jurisdiction, adding provisions that relate to how AI is used in the process of providing insurance coverage doesn't seem like something the Congress could legitimately intervene in.
This is going to be disastrous for hospitals and doctors, because they're facing a massive surge of (likely AI powered) denials and individual states are regulating it - this would ban that.<p>It's not like the laws prohibit any use of AI, it's literally basic safeguards and human in the loop provisions but the text of the bill as written would make those laws illegal.<p>Which is not surpsing considering it comes coupled with massive cuts in Medicaid - private Medicaid plans are some of the most egregious players in terms of denials.
> regulating artificial intelligence models, artificial intelligence systems, or automated decision systems<p>Seems trivial to work around since there is no legal definition of AI.<p>Instead of making your law specific to AI system, you can simply make it slightly broader in scope so it includes AI systems in practice.<p>For example, prohibition on AI facial recognition in public spaces -> prohibition on any computerized facial recognition
I suppose favoring "state's rights" over federal regulation is only a concern for the GOP when they're not getting big tech lobbyist money.
The people that are championing this sort of stuff, what's your take on social credit systems (like in China), or just total surveillance?<p>I'm asking, because my take is that totally unregulated AI will sooner or later lead to such applications. And you can't really advocate that privacy laws will stop that - after, that would hinder the progress of things like "automated decision systems".
if you ever wanted to obliterate any consumer confidence in a market thats already routinely mocked, loathed and derided...i can think of no better way than to ensure it is fecklessly unaccountable to any sort of regulation.
<i>What could possibly go wrong?</i> is no longer an exclusion but an enumeration I guess. Everything you can think of probably will. Could things like this be repealed when someone who knows what they’re doing steps in?
Since it's a reconciliation bill, is this likely to make it past the "Byrd bath"? It's looped in with a $500M AI modernization fund but my simplified understanding is that items not related to budget can be challenged and removed. Couldn't find reference to this in any of a few news articles.
I like how this is the exact opposite of what the EU's doing: <a href="https://artificialintelligenceact.eu/" rel="nofollow">https://artificialintelligenceact.eu/</a>
State and local governments cannot regulate. This means that the leader still can issue executive orders, e.g., against AI wokeness. Republicans very much wanted to regulate just one year ago:<p><a href="https://www.theguardian.com/us-news/2023/aug/21/artificial-intelligence-culture-war-woke-far-right" rel="nofollow">https://www.theguardian.com/us-news/2023/aug/21/artificial-i...</a>
Queue the usual remarks about "automated decision systems": is the PID controller in an espresso machine an automated decision system, is a pacemaker, Cochlear implant, fuzzy logic controller in a rice cooker, etc.