Max Winga

My biggest project yet is now LIVE! @hankgreen talks superintelligence, AI risk, and the many issues with AI today that present huge concerns as AI gets more powerful on SciShow. Happy Halloween! youtube.com/watch?v=90C3XV…

UK MPs were just selected to introduce bills of their choice. And we've published the updated version of our bill to ban superintelligence we presented to Number 10! If introduced, it'd be a world-first. Ask them to introduce it using our contact tool: ukbill.controlai.com

@jeffcafe_ It seems in fact bad for localized bans, because it just further decentralizes datacenters, making future prohibitions on ASI development harder to enforce.

@maxwinga Yes, that's important context about motivation. I've also made the same point about datacenter restrictions not working because they're so location-agnostic. Limited physical input/output and just needs energy (can be off-grid). Almost a fungible commodity.

Something many analysts are missing about the anti-datacenter push is that for people who hate AI for any number of reasons, fighting datacenters is the only lever they can see which hurts the AI companies. Unfortunately, even a national datacenter ban wouldn't actually prevent

@charliermarsh I'm Londonmaxxing harder (doing stuff on my computer, from London, and my name is Max)

I’m Londonmaxxing (doing stuff on my computer, but from London)

superintelligent AI and its risks. For that we need understanding of the situation and level of risk we face. It's very possible to prohibit superintelligent AI globally once it's common knowledge that anyone building it is akin to starting a nuclear war.

It's been lovely talking with @artchad and the CCCRU community. The risk posed by superintelligent AI impacts everyone, and I'm excited to see more creators grappling with it!

I'm a physicist with a background in AI safety research, I understand this technology and it's risks better than 99.99% of people. It's Really. Really. expensive to "just cut the wires bro" to the entire internet. Beyond money, lives will be lost, everything relies on the internet. Would you nuke a town to stop a mass shooter? The point of the kill-switch is to minimize the time it takes and the cost in the event you need to shut an AI down quickly. It's to establish the minimally invasive process.

AI companies are invested in people believing this bullshit. It's all science fiction bullshit people read and repeat without any logical thoughts what so ever. Fuck me, people really need to comment on technology they dont bother to understand first AI is not fucking magic. So what if data centers exist around the world? Do you have any idea how much work it takes to keep the internet online? Do you seriously fucking believe the internet is some magic fantasy land that just exist magically in your heart? No, it's a complex network barely held together by tens of thousands of people maintaining it at any given point. We had to build so much redundancy because it fails easily. The architecture of internet is incredibly vulnerable and all that AI rely on to move around can be easily shut down, data centers are easily isolated, and by cutting a few dozen cables, most of the world internet can be effectively cut off.

Oftentimes people will say about superintelligent AI "Why can't we just turn it off?" Well...we literally don't have a kill-switch right now. This amendment would require them, a common-sense measure!

.@Discoplomacy This kill-switch provision helps govt intervene in cases where UK datacenters are being maliciously used by a threat actor to conduct cyberattacks. It also helps in cases where the government has reason to believe superintelligent AI, AI that can autonomously compromise national security, is being developed on UK soil. While this doesn’t deal with all scenarios, it’s a common sense step forward. The UK is not truly sovereign on AI if it can't pull the plug when AI national security threats happen on its soil.

@GlenWilsonIA AI systems are increasingly integrated across the economy, running on hundreds of datacenters around the world. There's not just "a plug" that you can pull. AI companies themselves say their systems may eventually cause mass catastrophies, our leaders need a kill-switch.

@maxwinga It's called the power plug.... seriously. AI is not magic. A fucking squirrel knows how to kill an AI more than you, a supposed human, does.

Peter Jackson: "We're not talking about AI in general – the thing that's going to destroy the world..."

@Ahmadullah_Faiz ControlAI already exists @ControlAI, controlai.com

Just launched ControlAI. An app that helps you control how AI writes your code. Connect your GitHub repo, generate rules, and commit them instantly so tools like Claude, Cursor, and Codex stop writing messy, inconsistent code. No more random outputs. More predictable, structured code every time. This is just the MVP. Would love your honest feedback 👇 controlagent.io

@ilex_ulmus Nice, seems interesting! Some ideas: Should PauseAI lean in / out from the general anti-AI omnicause? What are intermediate goals PauseAI is pursuing that lead up to international pause agreements? How do you manage volunteers at scale? How do you moderate / filter members?

I’m doing a blogpost a day on s*bstack in June to explain my worldview with PauseAI. Subscribe so you don’t miss it. Also comments on for requests.

@hypertheoryalex @AndyMasley Sure, happy to meet you!

@maxwinga @AndyMasley Would be good too @maxwinga ; if you wanna meet up with EA Oxford they’ll be a strong contingent of us :)

If you're in London and wanna say hi between the 25th and 29th please let me know

@AndyMasley @hypertheoryalex Would love to have a chat at EAG!

@hypertheoryalex Yes I will!

Mythos seems to be a huge wake up call for NatSec people, and this is just a fraction of the raw capabilities of superintelligent AI. AI systems are increasingly becoming threat actors themselves, not just people misusing them...and that threat is getting bigger fast.

@MrEwanMorrison @spectator Happy to once it's no longer a threat

@spectator Oh pipe down about extinction

We are entering an era where the AI systems themselves are threats, not just humans ✍️ Andrea Miotti spectator.com/article/we-are…

When someone named Connor Johannes is warning you about superintelligent AI posing an extinction risk to humanity, you should listen!

@andreamiotti @spectator Great to see!

While AI leaders warn superintelligent AI could cause human extinction, governments are asleep at the wheel. The companies building superintelligence admit the danger, yet expect to create the tech within a few years. My piece in @spectator on the threat and what we can do.