Jason Calacanis’ Post

Transcript

The news over the weekend is that SB 104.7 in California was vetoed by Governor Gavin Newsom. I think we should explain maybe the top two or three concepts of how they were going to throttle monitor AI because there there are fears that people have and the fears should be real. The fear of job displacement is obvious. We're going to have on 10/10 the robo taxi from Tesla. We have BYOD, Waymo and four other companies partnering with Uber on automation, so the idea of being a taxi driver. Or a truck driver is going to radically change in our lifetime and it might change in I would say between 5 and 10 years depending on the jurisdiction. So. That's very real, what the government should do about that, if anything. I kind of wonder as a libertarian leaning kind of person, like with coal workers. I kind of feel like it's a small number of people that drive a lot of votes. It maybe, you know, in society we just say here's, you know, when, when these markets come apart, just here are some training maybe. And if you choose to take it, you know, here's five $10,000 worth of training. A nice thing a mitzvah for the government to do, but otherwise. You have to be responsible for your own future and see that coming and make changes in your career choices and or spending as an individual, right? Like this is part of living in a free dynamic society. That's a personal protectionist society. So anyway, let's go through the top two or three things. The fears are real. I don't think the fear of like they're going to come and kill us is real. I think that's like the 1% science fiction 1020 years from now. But I do think the job displacement is the one. That unions. And individuals should actually realistically care about well, that I think breakdown is why some people were opposed to ask you 104.7, because this is the bill all about catastrophic risk. And So what that means is what they call critical arms. It was talking about like avoiding AI systems helping create WMD that result in mash casualties or mass casualties and mass damage from cyber attacks. So essentially this was the bill that I was thinking about the far future, the biggest models and the biggest risks and that. This is how it was set up. So one of the reasons why this bill was so controversial is that it had a definition of essentially what models had to obey its rules. And those were ones that had computing that were built using computing power over 10 to the 26th integer or floating point operations or that cost $100 million or more to make. So Jason, you know, this bill wasn't about job protection. It wasn't about rescaling. It wasn't about UB. I it wasn't about the stuff that I think you and I really care about, which is people. It was. About trying to put some guardrails in place to prevent AI from the future, from becoming Terminator. And that's what I thought was kind of silly because it felt like it was waving at a future that we should maybe be afraid of and then trying to harm the president to get us there. And the way he did that was essentially adding civil and criminal penalties to companies that make AI models over a certain size if they did not follow the rules, preventing misuse of cover models, building kill switches. Filing lots of paperwork showing code listing where they got data and so forth. So there was a lot of regulations for these large models and now. No, Yeah. So if you were to regulate large models in California, in California is on the vanguard of creating regulations and creating new trends, I think California likes to try like. Regulations and getting up in people's business in a in a very big way. Texas, Florida live and let live. New Hampshire live free or die. There's like lots of different flavors in our country. And so California is going to California. And this is obviously like kind of ridiculous to. Create all this regulation. I know some people like probably Sam Altman and open an eye or Microsoft might actually want this. Because it creates $100 million a year in legal fees or $10 million a year in legal fees that will stifle other companies so or make it impossible for startups to compete because the legal bill would be greater than their team size and their staffing bill. So anyway, good riddance to this one. Europe will probably have 10 times worse. And really the focus I think that the government needs to look at is. What are we gonna do if we have Cataclysmic? Uh, job displacement. And you know what? We had cataclysmic job displacement multiple times in our history. Wars. Um, COVID stock market crashes and we have this concept of unemployment, we have this concept of stimulus that we drop people, you know, $1000 on people's heads like 3 or 4 times. So I'm not worried about it. If we have cataclysmic job displacement, we're a strong country with the ability to manage it. I, I do think it could create civil unrest though. What's funny is talking about this stuff. It's, it's how broad the impacts of AI could be. And therefore that's how scattershot the the AI regulations are. So when Gavin Newsom vetoed SB 104.7, he also kind of flexed that he had signed a lot of AI bills. I'm not going to go through them all by name and number her Jason to save everybody, but they were working on things in California like not using AI to create child sexual abuse material. Good, you know what I mean No cloning dead people without permission, pretty reasonable disclosing that an ad was a I edited reasonable. So there have been in California some nibbles at AI regulation, but to me they land on the side of use. Do not misuse this technology that has been built because that is the risk versus looking at the models themselves. And that to me was always the mismatch with this bill versus everything else because we can sit down and say, hey, let's not let teens make sexual deep fakes of one. Before about 8000, different reasons, right? No one's really going to be posted except for the most ******** libertarian, but there's like seven of those. So we don't need to worry about it too much. But to me, this, the idea of regulating at the model level was a big mistake. I think, by the way, that's like a really great point that I didn't even get to, which is, you know, Esther Dyson, who was a like a mentor of mine when I started in publication, she had a publication release 1.0. She always made time for me. Thank you, Esther, for that. You know, she said you don't have to use this technology. You could opt out of it if you disagree with what the company's doing. And you, you could not buy their product. So if you disagree with Seidel Hills or you disagree with Microsoft and how they're running things, there are alternatives. Vote with your dollar. And then the second thing she had said to me at some point was. Are we going to blame Volvo when a bank robber steals a Volvo and robs a bank? Is there like another place to put the blame as opposed to Volvo? And so you know that that analogy is a really good one until it isn't. You know, if somebody makes an AI drone specifically designed to target a person by photo and murder them by attaching like a gun to them, you know, OK, slightly different, but the drone and the gun, you know, that you use for hunting or that you use for making pictures, those people. Maybe you shouldn't be at fault if a lunatic puts a gun on a drum. I'll just end my Ted talk there. But I do think you're right and it is one of the arguments I had with Lena Kahn. I love when Lena Kahn takes action against specific things. The app stores are a good example of the tax in app stores or bundling. Those seem like really interesting. Tactical things to go after. Philosophical things, I think you get yourself in trouble, right? Because now you're kind of like philosophically, I want to stop technology and progress or philosophically, I don't believe people should be able to own a firearm. Like it just kind of shuts the the discussion down and that's why I wanted you to just break down tactically like you did wonderful here, founders, do you want to sell to bigger customers? I know you do. You got to get that ACV trending up and you want to push your chin down, right? Sounds good. But to sell to those big buyers, you need 2 clear. All of these compliance checks, you know that that means you got to have things like sock 2 sorted out what sock too? It's a standard and insures that companies keep their customer data safe. And if you are in sock too compliant, you can kiss those big deals goodbye. You're not going to land the lighthouse customers. You're not going to be able to operate at the highest end of the market, but Vanta makes it really easy for you to get and renew your sock 2 compliance. On average, Vanta customers or compliant in just two to four weeks can take months without Vanta and they automate. Clients for GDPR, HIPAA and more O you could sell Tabara customers in whatever market you're startup is going after. Vanta is going to save you hundreds of hours of work and U to 85% on compliance costs. Stop slowing your sales team down and use Vanta. Get $1000 off at vanta.com/twist. That's vanta.com/twist for $1000 off your sock too.