If OpenAI is So Worried That AI Will Kill Us, Let’s Get Weapons Inspectors Inside Its Labs!
It has come to my attention that OpenAI and a16z’s portfolio of companies, more broadly, are very concerned about its potential lethality, and inviting ideas on how to regulate these new technologies in order to prevent that future. While some might take this as a cynical play to manipulate regulatory boards (it is), OpenAI in particular has been very serious about telling us that they believe that the AI technology that they are developing, has the possibility of becoming lethal technologies that they could lose control of, and thus, it is incumbent on the global community, to work in partnership with OpenAI to develop processes and regulations so they don’t kill us all.
Thank you for informing us you are developing what you believe to be lethal technologies which you have little control over and don’t understand the risks of, and are now asking regulatory boards with no fucking CS training how to help you stop the apocalypse you are creating, at accelerationist speeds — the money behind OpenAI and other AI technologies, Marc Andreessen of a16z, openly identifies as an “AI accelerationist”. The confluence of these points does indicate to me that we are facing possible mass death events, not related to the characteristics of AI itself, but because the people developing it are showing time and time again that they are acting at the height of reckless endangerment, using illegal and stolen resources to build their tools, pushing AI into weapons startups, absolutely 0 engagement with the general public about this, no guarantees about it, no visibility into what is actually being built in the lab — not just the shiny applications and sand boxes they show us to monitor how we interact with the machine — no plan for how to prevent the worst futures, all while they literally integrate OpenAI with the actual factual Pentagon death machine, with both Anduril and Palantir, also part of the a16z portfolio, already integrating this into their lethal weapons systems as part of a comprehensive build out of a startup military with massive weapons contracts already.
Alright. You want regulation support? Let’s get some fucking international global weapons inspectors inside you. Especially because you all have been talking a wholeeee lot about building an atomic bomb. I feel like if that is what these people are doing, that is something that the international community needs to know about. They should also be put in front of Congress and other bodies to specifically talk about the fact that while OpenAI is speciously pointing to lethal threats of unknown quantities, it is also rapidly integrating with weapons platforms of known quantities (lethal drones, surveillance towers, in the kill chain, in new AI-enabled battlefield systems), which seems that that would be a far more immediate concern in the lethality question — and perhaps the actual lethality issue they are trying to avoid with this red herring.
Sam Altman in particular seems nervous about his AI killing us. I would first like to suggest that a technology executive with so little control over his own efforts, and so little conviction in his team, that he thinks lethality and an extinction event could emerge from it… that maybe we can get someone a little more competent in there. That said, assuming Sam is in good faith (lol), regarding his concerns, I think he brings up a great point which is simply: we do not know, and we need processes in place to help us prevent the worst outcomes. Since we are dealing with potential lethality, I suggest that startups who self-identify as posing a potential existential threat to humanity, submit to regular, independent, international weapons inspections will full access to all of the technology to assess the lethality threat and if necessary, shut down operations.
Weapons inspections helps us get in there and make sure that we have direct eyes and a direct assessment of the technology from an independent party. As of this moment there is literally zero auditing of technology done by VCs that aren’t over the course of their sales processes to customers including the CIA, and indeed, the vast majority of software development is done with literally no one outside the people in the startup directly working on it or having knowledge of it.
I think one thing this situation demands is a massive increase in public transparency about the development project and specifically relating to the roadmap, which should be openly published, so we have the ability to provide assessment about risks and intervene before the thing is actually coming out at the world with all of those risks already inside it, and a dopey, paranoid CEO raised by technofascists since he was 19, shrugging his shoulders to say: “I dunno. You decide. We think it might kill everyone”
Okay little bitch. Let’s see what’s in all your computers for real.
Weapons inspection I think is particularly important also in the context of, and thank you for pointing this out Sam, that now that VC is openly building an entire warfare platform, full of large weapons contractors and orbiting smaller startups serving various purposes, the question of inspection is particularly prescient. What does it mean that venture capital is now building weapons? Who outside of their customers knows what they are building? Who knows what it the furtherest they’ve gotten? What is in some of these labs? What is in the underground bunkers that literally all of the VCs have built on owned property all over the globe, where they are subject to the jurisdiction of the host country? Perhaps they would like to assist in getting into those and seeing whats actually in there. Has anyone ordered any plutonium lately? Uranium? Marc said on a recent podcast that obtaining these ingredients for a nuclear bomb is much easier than it used to be.
Indeed, VCs have a number of areas where they could develop different technologies in a total vacuum outside of the normal corporate offices. They have the money, power, resources, talent, knowledge, land and facilities, to have technological development happening outside of even the proper facilities. Areas I would suggest we should gain some transparency into include, the network of underground bunkers maintained by these weapons CEOs, inside of all the datacenters, in the private residences of tech billionaires where they have their own medical set-ups, computers and god knows what.
Like… how easy, exactly, Marc, is it to get the bomb stuff? And how do you even KNOW the ease of acquisition of nuclear bomb components? And why are you talking about the importance of building nuclear bombs in literally every podcast and blog post, including 20 minute long meditations on how Oppenheimer was a pussy because he went through a gauntlet of moral conscious in the development process? Many sociopaths and psychopaths do not experience such crises, in line with Marc’s total disdain of Oppenheimer’s lack of steely masculinity in the face of killing millions of children. Cool.
Open up the hood boys. Let’s see what we’re working with in here.
You got a bomb in there boys?