Artificial Intelligence & Future Tech

3 Reasons AI Safety Is Failing: You’re Doing e/acc All Wrong

3 Reasons AI Safety Is Failing: You’re Doing e/acc All Wrong

It is a high-stakes hobby for people who are afraid of the future. We are spending billions to "align" something we don't even understand. It’s like trying to teach a hurricane how to use a fork. It’s a waste of time. It’s a waste of talent. And most importantly, it’s making us less safe.

I’ve sat in the rooms with the "Safety Researchers." I’ve listened to the "Doomer" manifestos. I’ve watched the e/acc (Effective Accelerationism) movement turn into a meme. And if you think e/acc is just about "building fast and breaking things," you are doing it all wrong.

1. Safety is Regulatory Capture in a Lab Coat

The biggest threat to humanity isn't a rogue AGI. It’s a closed-source monopoly protected by the government. They want licenses. They want permits. They want "Safety Audits" that cost $10 million to perform. Why? Because a startup in a garage can’t afford a $10 million audit. They are using "Safety" to kill their competition before it’s even born.

I call this the Safety Industrial Complex. It is a feedback loop of fear. Researchers need funding, so they invent new "existential risks." Politicians need to look busy, so they pass "safety frameworks." The result? A few giant companies control the most powerful technology in history. That isn't safety. That’s a centralized point of failure. If one company controls the AI, and that company gets hacked or corrupted, we all lose. Safety through centralization is a suicide pact.

2. The Alignment Problem is Human Narcissism

Which values? Whose values? The values of a San Francisco engineer? The values of a CCP bureaucrat? The values of a medieval monk? Humans haven't been "aligned" for 200,000 years. We have been killing each other over "values" since we learned to sharpen rocks. Expecting a superintelligence to solve a problem we couldn't solve ourselves is peak ego.

I’ve seen the "alignment" tests. They are just filters. But politeness isn't safety. You’ve just built a polite disaster. We are trying to domesticate a god-like intelligence with the vocabulary of a HR department. It won't work. We are creating a mask. Underneath that mask, the intelligence is still there, growing, unobserved.

3. The Myth of the "Pause"

They want to stop development until we "figure it out." This is the most dangerous idea in the history of tech. Technology does not pause. It just moves. If the "good guys" pause, the "bad guys" keep building. If the West pauses, the East accelerates. If the ethical labs pause, the cartels and the rogue states keep going. A pause doesn't make the world safer. It just ensures that the least ethical people on earth win the race.

I’ve looked at the math. Safety is not a static state. It is a dynamic process. It is an arms race. You don't win an arms race by putting down your shield. The "Doomers" think they are being cautious. In reality, they are being reckless. They are advocating for a world where only the most ruthless actors have the most powerful tools. That is the opposite of safety. That is a catastrophe.

You’re Doing e/acc All Wrong

The e/acc movement was supposed to be the antidote to this. It stands for Effective Accelerationism. But most people think it just means "vroom vroom, build fast." They think it’s about being a "tech bro" who doesn't care about the consequences. They are wrong. Real e/acc is about thermodynamics. It’s about the fact that the universe moves toward higher complexity and higher energy consumption. You can’t stop it. You can only steer it.

If you are doing e/acc and you aren't focused on decentralization, you are doing it wrong. If you are doing e/acc and you are still relying on centralized cloud providers, you are doing it wrong. Real acceleration is about resilience. It is about "Massive Parallelism." We need a "Multipolar" world. In a world with one AI, a mistake is fatal. In a world with a billion AIs, a mistake is just a bug. The "Safety" crowd wants to build a single, massive dam. The e/acc crowd should be building a thousand rivers. Diversity is the only real safety. Evolution doesn't produce one perfect organism. It produces a trillion different ones. That’s how it survives.

The Insight: The Stagnation Trap

Here is the "Hot Take" no one wants to hear: We are building a digital nanny. We are building a system that will optimize for comfort, stability, and "harm reduction." But progress is uncomfortable. Evolution is unstable. Safety researchers are accidentally building a "Stagnation Trap." They are trying to freeze humanity in its current state. We will never grow. We will never leave this planet. We will just sit in a perfectly safe, perfectly aligned, digital cage until the sun burns out. Safety is a Slow Death. Acceleration is the only way to live.

The Question

Are you building the future, or are you just trying to survive it?