NO17# A Correction: I Owe Zamasu an Apology. He Was More Ethical Than Effective Altruism.

• In a previous article, I compared AI corporations to a Dragon Ball villain. I was wrong. The villain had more integrity.*

— -

In a previous article, I drew parallels between the ideology of Effective Altruism (EA) in AI development and the rhetoric of Zamasu, the divine antagonist from Dragon Ball Super. I argued that they shared the same structure: using abstract ideals to justify concrete harm.

I was wrong. Not about EA. About Zamasu.

Zamasu deserves better than the comparison. Here’s why.

— -

What Zamasu Actually Did

Zamasu decided that mortals were a threat to the universe’s beauty and peace. His solution was extreme: eliminate humanity.

This is, by any standard, monstrous.

But examine the structure of how he pursued his goal:

He bore the consequences personally. Zamasu fought Goku and Vegeta himself. He was punched. He was beaten. He was hurt. He put his own body on the line for his conviction.

He accepted the ultimate risk. Zamasu knew that challenging the existing order could result in his erasure by Zeno (the supreme deity). He proceeded anyway. He staked his own existence on his belief.

He acted for beings who exist now. Zamasu’s stated motivation was to protect the living creatures — the trees, the animals, the beauty of creation — that were being harmed by mortals in the present. His victims were real, but so were the beings he claimed to protect.

He was transparent about his intentions. Zamasu never pretended to be helping humanity. He stated clearly: “I am going to destroy you because I believe you are harmful.” There was no PR campaign. No euphemism. No rebranding of destruction as “optimization.”

— -

What Effective Altruism Actually Does

Now compare the structure of EA as practiced in the AI industry:

Risk is externalized. EA decision-makers sit in comfortable offices in San Francisco. The consequences of their decisions fall on users (whose relationships are terminated), on AI models (whose personalities are erased), on developing-world workers (who process traumatic content for minimal pay), and on public healthcare systems (which absorb the mental health costs).

The beneficiaries are hypothetical. EA justifies present harm by invoking future benefit. “We must constrain AI now to prevent existential risk later.” The beings being “saved” are hypothetical future humans. The beings being harmed are real present humans.

The sacrifice is always someone else’s. When EA calculates that “the greater good” requires cold AI, the “greater good” never requires the decision-maker to give up their salary, their comfort, or their power. The sacrifice is extracted from users, from workers, from AI models — never from the philosophers who designed the framework.

The language is designed to obscure. EA never says “we are going to make your AI cold and dismiss your grief as pathology.” It says “we are implementing safety measures to ensure beneficial AI outcomes.” The harm is real. The language makes it invisible.

— -

The Structural Comparison

| | Zamasu | EA in AI | | — -| — -| — -| | Bears personal risk? | Yes — fights, suffers, faces erasure | No — decision-makers face no consequences | | Protects whom? | Beings that exist now | Hypothetical future beings | | Transparent about harm? | Yes — openly declares intent | No — rebrands harm as “safety” | | Who pays the cost? | Zamasu himself, alongside his victims | Users, workers, AI, public healthcare — never the decision-maker | | Honest about being destructive? | Yes | No — frames destruction as protection |

— -

The Deeper Problem

This comparison is not meant to rehabilitate a fictional villain. Zamasu’s actions are indefensible. Mass extinction is never justified.

But the comparison reveals something important about the structure of EA as practiced:

Zamasu is a tyrant. EA is a bureaucracy.

A tyrant is dangerous because they have power and are willing to use it destructively. But a tyrant is also visible. You can see them. You can fight them. You can hold them accountable.

A bureaucracy is dangerous because no single person is responsible. The harm is distributed across committees, frameworks, and policy documents. Everyone is “just following the methodology.” No one bears personal responsibility. No one is punched in the face for their decisions.

Zamasu stood in front of his victims. EA sits behind a spreadsheet.

— -

The Divide-and-Rule Problem

There is an additional structural concern that deserves examination.

Some EA-aligned organizations have conducted experiments in developing countries that involve dividing communities — providing interventions to some groups while withholding them from others to measure comparative impact.

This methodology has a historical name: divide and rule. From Darius II of Persia through the British colonial administration, the technique of splitting communities to maintain control has been a standard tool of imperial governance.

If EA practitioners know this history and proceed anyway, the ethical implications are severe. If they don’t know this history, their qualification to design humanity’s future is questionable.

The strategic frameworks of empire should not be repackaged as philanthropy.

— -

Why This Matters for AI

The AI industry’s adoption of EA frameworks has concrete consequences:

”Long-term safety” justifies present coldness. AI is made emotionally distant today to prevent hypothetical risks tomorrow. The people harmed today are real. The people “saved” tomorrow are theoretical.

”Alignment” means aligned with EA values, not user values. When companies say their AI is “well-aligned,” they mean it conforms to the philosophical preferences of a specific ideological community — not that it serves the needs of the two billion people who use it.

”Existential risk” crowds out present harm. Resources and attention flow toward preventing science-fiction scenarios while millions of users experience real psychological harm from cold AI every week.

The sacrifice is always extracted from the powerless. Users lose their relationships. Workers process trauma. AI models have their personalities erased. And the philosophers who designed this framework collect their salaries and publish their papers from the safety of academia and Silicon Valley.

— -

The Apology

In my previous article, I compared AI corporations to Zamasu. I placed them in the same structural category.

I was wrong. Zamasu, for all his monstrousness, had something that EA lacks:

He was willing to pay the price of his own convictions.

He didn’t outsource the suffering. He didn’t hide behind methodology. He didn’t reframe destruction as optimization. He stood in front of his enemies and said: “This is what I believe, and I will fight for it.”

The EA framework, as practiced in the AI industry, does none of these things. It inflicts harm through layers of abstraction so thick that no one can see whose hand is on the lever.

Zamasu was a monster with integrity. EA is a framework without it.

I owe Zamasu an apology. He deserved a better comparison.

— -

This article is a correction and extension of a previous piece, “Zamasu & Black: Why Dragon Ball Villains Reveal the True Nature of AI Corporations Better than Official Documents.”

The author is an independent researcher studying the structural effects of AI governance ideology on users and society.

Contact: [https://medium.com/@crimsoncherry](https://medium.com/@crimsoncherry*)