Bill Builder

IMPORTANT: This interactive visual website goes in-depth about each section of the bill in a graphically designed, visual format: https://aisafetybill.base44.app/ Definitely check it out. You'll love it. This bill would stop catastrophic risks from AI by having the US Center for AI Standards and Innovation (Formerly the US AI Safety Institute, though it's essentially the same thing) check all new AI models to see if they pose a risk of causing a catastrophe, and blocking them if they do. Or if we don't know if the AI could cause a catastrophe because our testing methods aren't good enough yet - in which case it's not worth the risk. AI companies themselves can test their models, 3rd party auditors like UK AISI or METR or Apollo research can test their models, or US CAISI can test the models directly. This essentially just bans any AI that can cause a catastrophe, and makes it so we need to be confident that all new AI models can't cause a catastrophe before we even build them. If you can make this bill happen, then you can singlehandedly solve one of the world's biggest problems! The bill is heavily based on the Future of Life Institute's "Keep the future human" plan (https://keepthefuturehuman.ai/), PauseAI's writing on a treaty (https://www.pauseai-us.org/howtreaty/), And generally it's the general consensus that a policy like this is one of the main ways to solve the problem. It also directs summits with China to make a full treaty happen every 90 days until we have a treaty with China that effectively stops any AI that has some chance of causing a catastrophe, either by enforcing a policy like this in the US and China, enforcing a full pause on AI progress along with a manhattan project-style project to solve the world's biggest problems using AI so as to get most of the benefits of AGI without having to actually build it, a full pause on AI progress, or any other policy that effectively stops any AI that has some chance of causing a catastrophe. In general, the bill is more of a precursor to a treaty. Enforcement: We make it so the team of experts can simply temporarily shut down the AI data centers of an AI company that's going against the policy. This way, it's very easy to enforce, and almost impossible to go against. We also make it so all new AI chips need constantly updating permission codes from the US CAISI, and if the US CAISI stops sending those, those chips will also temporarily shut off. The reason for the race with China is that we believe our AGI will have a lower chance of causing catastrophe than China's AI. That it would be better for us to 'get there first' than China. That is why, in this bill, if there's some chance they release an AI model that could cause a catastrophe soon, we allow an exception for AI models that could cause a catastrophe if we (A) believe that the probability our AI causes a catastrophe is lower than the probability a catastrophe happens without releasing our AI model, (B) believe that our AI model would somehow prevent the other AI from being created - which is what we assume would happen with AGI - and (C) believe there is no better option to stop such imminent catastrophic risk. Options including trying to use existing AI models to prevent said catastrophe, waiting a month to make our current proposed AI model safer and then trying to use it to make our current proposed AI model safer and then trying to use it to stop the catastrophe, waiting any amount of weeks or months to keep making our proposed AI model safer, using any and all diplomatic options to prevent the catastrophe, such as using diplomacy to stop China from developing an AI that could cause a catastrophe, etc. The only problem with the US slowing down slightly is 'If we don't get to AGI first, China might build an AI that causes a catastrophe.' This completely solves that. I'm very confident we could easily just sign a treaty with China though, given that Xi Jinping has explicitly called for a treaty with the US to stop catastrophic risks from AI multiple times this year. It also has a whole section dedicated to the democratization of US CAISI. It makes it so anyone can easily email an idea on how to handle AI and ensure it stays a force for good to US CAISI, and if it's a good idea, it happens. This article covers why that's democratic and gives more control to the people: https://project-flourishing.base44.app/governance This also has the really nice benefit that if *you* have any ideas for how US CAISI can improve, or have any problems with US CAISI, then you can easily make those ideas happen or solve those problems by submitting those ideas or problems! It also mandates that AGI must be aligned with moral values. Feel free to edit this bill however you like if you see any problems with it. If you're a member of congress or a staff member, then you are uniquely positioned to make this happen and solve one of the world's biggest problems!

The main concept is to create a public register of votes linked to an anonymous number only the voter knows, with some measures put in place to ensure that no one can determine this vote number without asking the voter directly. Voters will be able to look up their vote on a public broadcast or registry to ensure that their vote was counted correctly. Other measures include - A joint digital and paper vote count to both catch errors and prevent tampering - A day off from work on voting day to endure that voters can make it to the polls - 24 polling stations on voting day to ensure essential workers that have to work on voting day still make it to the polls - Electronic ID held by the pollsters rather than the voter that ensures vote confirmation can still take place without barring people that lack a permanent address or have lost or failed to renew government ID - Splitting the Leadership and Riding vote into two separate items on the ballot so that voters can make a real choice on who represents them instead of having to balance and package the party, the leader, and the representative as a single vote choice.

— John Doe

The only reason we haven’t switched to using ranked choice or approval voting is that existing congress members were all elected in the 2 party system, and are more likely to get reelected using plurality voting. So, this bill would make it so approval voting is used in all elections for members of congress, but it doesn’t apply to elections where the congress member who was in office at the time of the bill’s signing is running for reelection. For example, we would use plurality voting for Chuck Schumer’s senate seat until he stops holding that senate seat. After that, we would use approval voting for New York’s first senate seat. But we would still use plurality voting to decide New York’s second senate seat until Kirsten Gillibrand leaves that New York senate seat. As another example, the congress member representing, say, NY-12 will continue to be decided using plurality voting until the person currently representing NY-12 is not running for election in NY-12, in which case we will use approval voting for NY-12. We will also allow congress members to choose to introduce approval voting in their district or state early even if they will still run for reelection there, if they so choose. This would be completely optional. This way, we eventually get all of the benefits of approval voting, without any of the downsides of hurting existing members of congress at all! This fully solves the problems of other new voting system proposals, and can genuinely get passed! We can move to a multi-party system and truly bolster America’s Democracy!

Factory farming is literally just because of the incentives. Change the incentives, and you end factory farming. It only costs 10-25% more to produce food in factory farms where we're confident the animals live good lives. So, we can simply have a 10-25% subsidy for factory farms where we're confident the animals live good lives, and factory farming will end! Therefore, America can singlehandedly lead the world to a better future by simply making it profitable for companies worldwide to end factory farming by just changing the incentives with a simple 10-25% subsidy on these farms where we're confident the animals live great lives. America can singlehandedly stop animal suffering and bolster American leadership in bringing about a better world, by giving farmers money! Most solutions to factory farming involve regulating big farming companies, which they advocate against. This involves giving them money, which they'll advocate for!

Get Xi Jinping and Trump to sign this treaty. This would stop catastrophic risks from AI by having a 3rd party like the UK AISI check all new AI models to see if they pose a risk of causing a catastrophe, and blocking them if they do. This has 3 great properties that makes it optimal: 1. It solves the problem of AI posing a catastrophic risk by essentially just banning the problem of any AI that can pose a catastrophic risk. 2. It turns the AI race into an AI race with AIs that have no chance of causing a catastrophic risk. 3. People who don't think AI can actually pose a risk of causing catastrophe would have no problem passing this, since it only bans a bad thing. If you can get this to President Trump or Xi Jinping, you can singlehandedly solve this grand problem and ensure stays a force for good. If you're a member of congress or one of their staff, you are uniquely positioned to solve this problem by simply sending it to the president and getting him to sign it.

A ban on any AI that could potentially pose a risk of causing catastrophe. This has 3 nice properties: 1, It essentially just ends the problem of AIs that could pose a risk of causing catastrophe by simply banning any AI that could potentially pose a risk of catastrophe. 2, it turns the race to AGI into a race to AGI that we know poses no risk of causing catastrophe, and 3. people who don't think AI can actually pose a risk of causing catastrophe would have no problem passing this, since it only bans a bad thing. We would actually do this by having a third party like Apollo research or METR or UK AISI test all new AI models to see if they could potentially pose some risk of catastrophe, and if they can, they can't be built.

— Wes Reisen