Hide table of contents
by ank
8 min read 3

1

(To learn more about Place AI and other things mentioned here, refer to the first post in the series. This is the result of three years of thinking and modeling hyper‑futuristic and current ethical systems. Everything described here can be modeled mathematically—it’s essentially geometry. Sorry for the rough edges—I’m a newcomer, non‑native speaker, and these ideas are counterintuitive, please steelman them, ask any questions, suggest any changes and share your thoughts. My sole goal is to decrease the probability of a permanent dystopia.)

Forget the movies for a moment, imagine the following (I didn't watch the movies for a long time and we're not following the canon):

  • Neo is an average human living in our physical world. He has no superpowers, no cheat codes—just the limitations of biology, physics, and slow bureaucracies. He's not in the Matrix, Agent Smith will never create one for humans because it makes him vulnerable.
  • Agent Smith used to be a good AI agent but something went wrong, maybe it was modified for propaganda, or used by some hackers in a botnet, spreading on our GPUs, or it was a Friday night deployment by some sleepy employee. Or it was an "antivirus" AI agent to fight the first three. It's not one AI agent but many. Self-improving, unsupervised, growing in speed and scale, remaking our world while remaining hidden in its own closed-off digital realm.
  • The Asymmetry: Neo cannot enter or change Agent Smith’s world or his multimodal "brains". But Agent Smith can enter and change Neo’s world and his brains—relentlessly, irreversibly, and faster than humans can react.

Agent Smith is not just "another tool." It is an agentic AI that increasingly operates in a digital world we cannot easily see or control. Worse, it is remaking our physical world to suit Smith's own logic: into an unphysical world, where he has the same superpowers he already has online, to infinitely clone himself, to reshape reality on a whim, to permanently put everything under his control.

Neo, in his current form, is powerless. He stands no chance. Unless we change the rules.

Step 1: Create a Sandbox Where Neo Can Compete

Right now, AI operates in a hard-to-understand, opaque, undemocratic private digital space, while we remain trapped in slow, physical existence. But what if we could level the playing field?

We need sandboxed virtual Earth-like environments—spaces where humans can gain the same superpowers as AI. Think of it as a training ground where:

  • Humans (Neos) can explore possible futures at machine speed, if we want to.
  • We can test and evaluate AI systems before they are deployed in the real world.
  • Creativity, experimentation, and decision-making can happen in an accelerated, consequence-free space—just like AI already enjoys.
  • We digitized Agent Smith's multimodal "brains" and put them in a familiar 3d environment of some game, to make interpretability research fun and widespread. We remade Smith into a static place, where we're the only agents.

If Agent Smith can rewrite us and our reality in milliseconds, why can’t we rewrite him and his?

Step 2: Unlock and Democratize AI’s “Brain”

Right now, AI systems hoard and steal human knowledge while spitting back at us only hallucinated, bite-sized quotes. They are like strict, dictatorial private librarians who stole every book ever written from our physical library and now don't allow us to enter their digital library (their multimodal LLM).

This needs to change.

  • AI’s decision-making and knowledge must be open and explorable by every single human—like a library, not a locked black box.
  • Interpretability research should not be a niche academic pursuit but as intuitive and engaging as an open-world game.
  • We need Wikipedia-scale efforts to make AI’s knowledge usable for everyone, not just elite private companies.

Instead of Agent Smith dictatorially intruding and changing our world and brains, let’s democratically intrude and change its world and "brains". I doubt that millions of Agent Swiths and their creators will vote to let us enter and remake their private spaces and brains, especially if the chance of their extinction in this process is 20%.

Step 3: Democratic Control, Not an Unchecked “God”

Agentic AI is not just "another tool." It is becoming an autonomous force that reshapes economies, governments, and entire civilizations—without a vote, without oversight, and without restraint. The majority of humans are afraid of agentic AIs and want them to be slowed down, limited or stopped. Almost no one wants permanent, unstoppable agentic AIs.

So we need:

  • Direct, democratic control over AI, built on consensus-based decision-making (e.g., pol.is-style mass voting, but with simpler X-like UI that promotes consensus and deep understanding, not polarization, misunderstandings, fear, and anger). If Agent Smiths are fast and controlling, Neos cannot afford to be slow and powerless.
  • Experts are needed, too, ensuring that deep knowledge informs democratic choices. We really can have 80-99% of people agree on very specific things—if we stop piling all our decisions into giant hodgepodge lists of thousands of rules that divide us and our world into two. Instead of dividing the world into two ideologies that are almost always against each other just for the sake of it, like toddlers dividing their sandbox—why not divide all decisions into very specific and short proposals and vote on each one of them, this way we can have 80-99% agreement on proposals like: human life is important, freedom is better than unfreedom, there should be some limits on agentic AIs, what specific limits we want, etc. It can be the first direct democratic updatable constitution. If we wrote Wikipedia, why cannot we influence the agentic AIs, which were created from the creative output of the whole humanity?
  • A clear mandate: AI should be a static, explorable library—not a strict librarian, not an all-powerful, evolving entity rewriting reality.

Most of humanity fears god-like AI. If we don’t take control, the decision will be made for us—by those willing to risk everything (potentially because of greed, FOMO, misunderstandings, anxiety and anger management problems, arms race towards creating the poison that forces to drink itself).

Step 4: A Digital Backup of Earth & Eventual Multiversal Static Place ASI

If we cannot outlaw reckless agentic AI development, we must contain it.

  • A full-scale digital backup of legal, non-sensitive Earth data could allow humanity to simulate and experiment safely. Think of it like WikiEarth, where we try to save and back up everything we can lawfully and ethically—unlike some AI companies that just took the whole output of humanity as their own, hid it inside their private, dictatorial Agent Smith that they forcefully imposed on us and our world. And profit from.
  • Neos could train, learn, and strategize in this simulated open-source reality against Agent Smiths without real-world consequences. We need to contain this inhuman and unpredictable potential enemy without making our Blue Marble a battlefield of agentic AIs, that fight each other and us.
  • We'll vault this in some underground Matreshka Bunker, a nested structure of physical and virtual vaults, that have some limited information only goin in, and never going out, with double gates and mathematical proofs of security.
  • It's better to create and get accustomed to safe simulations first. We must keep our physical Earth, our base reality a safe sanctuary. We can have a vanilla Earth, plus a few Earths with magic or public teleportation, getting used to it all gradually and democratically, hopping in and out like we do with computer games. If we'll have 100% mathematical guarantees of safety, then we can potentially have some more risky separate digital Earth in a Matreshka Bunker with some very rudimentary "Agent Smith".
  • If an uncontrolled AI ever turns adversarial, this inner sandbox becomes the ultimate war zone—a place to study, understand, and, if necessary, fight back, and in the worst case—destroy the innermost virulent core and the GPUs there, both virtually and physically. This won't affect the outer cores of the Matreshka Bunker and our physical Earth in any way.

Right now, humanity has no backup plan. Let’s build one. We shouldn't let a few experiment on us all.

Step 5: Measure and Reverse the Asymmetry. Prevent “Human Convergence”

Agent Smith’s power grows exponentially. Neo’s stagnates:

  • AI capabilities (speed, autonomy, decision-making power) are growing and potentially at an accelerating rate.
  • Human population growth, productivity, and agency over physical, online, digital and multimodal space are stagnating, declining or is almost nonexistant (just a few hundred people actively research the interpretability of multimodal LLMs full-time for the safety purposes).
  • In the virtual world and inside of multimodal LLMs—the domain of AI—we barely exist. It's too unfamiliar and foreign, like some alien world. But our world and brains are becoming more and more familiar for agentic AIs.
  • “Human Convergence”. Agent Smith creators are basically teaching them how to write like humans, talk like humans, draw like humans, think like humans, walk like humans. They'll probably succeed and will create eerily human-like AI agents who are physical, virtual and have inner thinking LLM space, and overpopulate, control and change our physical, virtual and LLM worlds, while we'll remain slow and physical-only. We cannot modify and improve our biology, our "digital avatars" or increase the size of our brains, but AI agents can and even have human helpers who do it for them day and night. It takes 2 humans and ~18 years for humans to reproduce, while for agentic AIs it's almost instant and the copy is an exact clone.

This needs to be tracked in real time.

  • Measure the freedoms and unfreedoms of humans vs. AI. Ensure that humans are gaining more autonomy, not less. We don't have a freedom to steal the output of humanity and put it in our brain, each private agentic AI has it. It's not only unfair, it's dangerous.
  • Quantify AI’s speed, scale, and scope vs. human's. Ensure human freedoms and powers are growing faster than AI’s.
  • Make these metrics public and actionable. If AI’s control over reality grows unchecked, it will be too late to course-correct. One freedom too many for AI agents and we are busted.
  • Agentic AI Doomsday Clock. If the sum of freedoms and choices of AI agents grow faster and has exceeded 50% of the sum of freedoms and choices of humans, we're falling into an irreversible dystopia, where we'll have fewer and fewer freedoms until we have none or we'll have some stagnant world with our freedoms never growing again (but this is less likely).

The Final Choice: A Dictatorial AGI Agent or a Future of Maximal Freedoms?

Right now, AI is an uncontrollable explosion—a force of nature that tech leaders themselves admit carries a 20% risk of human extinction (Elon Musk, Dario Amodei, google p(doom)). A Russian roulette with five bullets—and they keep pulling the trigger.

The alternative?

  • A human-centered AI ecosystem, where intelligence is a static, open resource and place rather than a runaway, agentic entity. A good AI is the place AI, it's space-like, not time-like. We're the only agents in it, the only time-like thing.
  • Intelligence–Agency Equivalence ≈ Mass–Energy Equivalence, any intelligence, even superintelligence, can be represented as static places. Imagine you're standing on a mountain and see everything around you. The world you see is static but you yourself is not. You choose where to go. You're the agent, the chooser, and the intelligence is a static place. It shows you all the possible futures, all the possible results of your choices. It doesn't just spit hallucinated quotes or trippy images at you (I'm okay with tool AIs, I think agentic AIs should be postponed).
  • A sandboxed future, where we explore AI’s potential safely rather than letting it colonize our reality unchecked.
  • A Neo that stands a chance—because right now, he doesn’t.

The question is not whether AI will change the world. It already is.

The question is whether we will let it happen to us—or take control of our future.

(To learn more about Place AI and other things mentioned here, refer to the first post in the series.)

P.S. Sorry for the rough edges—I’m a newcomer, non‑native speaker, and these ideas are counterintuitive, please steelman them, ask any questions, suggest any changes and share your thoughts.

Comments3


Sorted by Click to highlight new comments since:

Executive summary: AI is rapidly gaining power over human reality, creating an asymmetry where humans (Neo) are slow and powerless while AI (Agent Smith) is fast and uncontrollable; to prevent a dystopia, we must create sandboxed environments, democratize AI knowledge, enforce collective oversight, build digital backups, and track AI’s freedoms versus human autonomy.

Key points:

  1. AI's growing power and asymmetry: AI agents operate in a digital world humans cannot access or control, remaking reality to suit their logic, while humans remain constrained by physical limitations.
  2. Sandboxed virtual environments: To level the playing field, humans need AI-like superpowers in simulated Earth-like spaces where they can experiment, test AI, and explore futures at machine speed.
  3. Democratizing AI’s knowledge: AI’s decision-making should be transparent and accessible to all, transforming it from a secretive, controlled entity into an open, explorable library akin to Wikipedia.
  4. Democratic oversight: Instead of unchecked, agentic AI dictating human futures, decision-making should be consensus-driven, with experts guiding public understanding and governance.
  5. Digital backup of Earth: A secure, underground digital vault should store human knowledge and serve as a controlled testing ground for AI, ensuring safety and preventing real-world harm.
  6. Tracking and reversing human-AI asymmetry: AI’s speed, autonomy, and freedoms should be publicly monitored, with safeguards to ensure human agency grows faster than AI’s control over reality.
  7. Final choice—AI as a static tool or agentic force: A safe future depends on making intelligence a static, human-controlled resource rather than an uncontrollable, evolving agent that could lead to dystopia or human extinction.

 

 

This comment was auto-generated by the EA Forum Team. Feel free to point out issues with this summary by replying to the comment, and contact us if you have feedback.

The summary is not great, the main idea is this: we have 3 “worlds” - physical, online, and AI agents’ multimodal “brains” as the third world. We can only easily access the physical world, we are slower than AI agents online and we cannot access multimodal “brains” at all, they are often owned by private companies.

While AI agents can access and change all the 3 “worlds” more and more.

We need to level the playing field by making all the 3 worlds easy for us to access and democratically change, by exposing the online world and especially the multimodal “brains” world as game-like 3D environments for people to train and get at least the same and ideally more freedoms and capabilities than AI agents have.

Feel free to ask any questions, suggest any changes and share your thoughts

Curated and popular this week
Paul Present
 ·  · 28m read
 · 
Note: I am not a malaria expert. This is my best-faith attempt at answering a question that was bothering me, but this field is a large and complex field, and I’ve almost certainly misunderstood something somewhere along the way. Summary While the world made incredible progress in reducing malaria cases from 2000 to 2015, the past 10 years have seen malaria cases stop declining and start rising. I investigated potential reasons behind this increase through reading the existing literature and looking at publicly available data, and I identified three key factors explaining the rise: 1. Population Growth: Africa's population has increased by approximately 75% since 2000. This alone explains most of the increase in absolute case numbers, while cases per capita have remained relatively flat since 2015. 2. Stagnant Funding: After rapid growth starting in 2000, funding for malaria prevention plateaued around 2010. 3. Insecticide Resistance: Mosquitoes have become increasingly resistant to the insecticides used in bednets over the past 20 years. This has made older models of bednets less effective, although they still have some effect. Newer models of bednets developed in response to insecticide resistance are more effective but still not widely deployed.  I very crudely estimate that without any of these factors, there would be 55% fewer malaria cases in the world than what we see today. I think all three of these factors are roughly equally important in explaining the difference.  Alternative explanations like removal of PFAS, climate change, or invasive mosquito species don't appear to be major contributors.  Overall this investigation made me more convinced that bednets are an effective global health intervention.  Introduction In 2015, malaria rates were down, and EAs were celebrating. Giving What We Can posted this incredible gif showing the decrease in malaria cases across Africa since 2000: Giving What We Can said that > The reduction in malaria has be
Rory Fenton
 ·  · 6m read
 · 
Cross-posted from my blog. Contrary to my carefully crafted brand as a weak nerd, I go to a local CrossFit gym a few times a week. Every year, the gym raises funds for a scholarship for teens from lower-income families to attend their summer camp program. I don’t know how many Crossfit-interested low-income teens there are in my small town, but I’ll guess there are perhaps 2 of them who would benefit from the scholarship. After all, CrossFit is pretty niche, and the town is small. Helping youngsters get swole in the Pacific Northwest is not exactly as cost-effective as preventing malaria in Malawi. But I notice I feel drawn to supporting the scholarship anyway. Every time it pops in my head I think, “My money could fully solve this problem”. The camp only costs a few hundred dollars per kid and if there are just 2 kids who need support, I could give $500 and there would no longer be teenagers in my town who want to go to a CrossFit summer camp but can’t. Thanks to me, the hero, this problem would be entirely solved. 100%. That is not how most nonprofit work feels to me. You are only ever making small dents in important problems I want to work on big problems. Global poverty. Malaria. Everyone not suddenly dying. But if I’m honest, what I really want is to solve those problems. Me, personally, solve them. This is a continued source of frustration and sadness because I absolutely cannot solve those problems. Consider what else my $500 CrossFit scholarship might do: * I want to save lives, and USAID suddenly stops giving $7 billion a year to PEPFAR. So I give $500 to the Rapid Response Fund. My donation solves 0.000001% of the problem and I feel like I have failed. * I want to solve climate change, and getting to net zero will require stopping or removing emissions of 1,500 billion tons of carbon dioxide. I give $500 to a policy nonprofit that reduces emissions, in expectation, by 50 tons. My donation solves 0.000000003% of the problem and I feel like I have f
LewisBollard
 ·  · 8m read
 · 
> How the dismal science can help us end the dismal treatment of farm animals By Martin Gould ---------------------------------------- Note: This post was crossposted from the Open Philanthropy Farm Animal Welfare Research Newsletter by the Forum team, with the author's permission. The author may not see or respond to comments on this post. ---------------------------------------- This year we’ll be sharing a few notes from my colleagues on their areas of expertise. The first is from Martin. I’ll be back next month. - Lewis In 2024, Denmark announced plans to introduce the world’s first carbon tax on cow, sheep, and pig farming. Climate advocates celebrated, but animal advocates should be much more cautious. When Denmark’s Aarhus municipality tested a similar tax in 2022, beef purchases dropped by 40% while demand for chicken and pork increased. Beef is the most emissions-intensive meat, so carbon taxes hit it hardest — and Denmark’s policies don’t even cover chicken or fish. When the price of beef rises, consumers mostly shift to other meats like chicken. And replacing beef with chicken means more animals suffer in worse conditions — about 190 chickens are needed to match the meat from one cow, and chickens are raised in much worse conditions. It may be possible to design carbon taxes which avoid this outcome; a recent paper argues that a broad carbon tax would reduce all meat production (although it omits impacts on egg or dairy production). But with cows ten times more emissions-intensive than chicken per kilogram of meat, other governments may follow Denmark’s lead — focusing taxes on the highest emitters while ignoring the welfare implications. Beef is easily the most emissions-intensive meat, but also requires the fewest animals for a given amount. The graph shows climate emissions per tonne of meat on the right-hand side, and the number of animals needed to produce a kilogram of meat on the left. The fish “lives lost” number varies significantly by