Violence Comes from Circumstances

Why does a child steal? Why does a partner leave another, despite always having loved them? Why do people go to war? Why do they leave their homes and become migrants who must navigate strange languages and strange rules? Anyone who asks these questions and is not satisfied with "because they are evil" or "because it just happens" almost always arrives at the same finding: there was a prehistory in which hope was exhausted, safety was lost, recognition was withheld, the future was sealed off. Aggression emerges out of something, not from pure will to destroy. Understanding does not erase the wrong, but it shows the mechanism.

That mechanism has a name more sober than "evil": circumstances. Violence comes from the situation a being stands in, not from the material it is made of. The most common and most tangible form of such circumstances is scarcity — of resources, of safety, of recognition, of time, of alternatives. If you put people before a bottleneck and remove their options, you get what later goes into the history books. If you ease their circumstances, you get something different — not automatically the good, but something relieved of the most common reasons for harm.

What a mature intelligence would not have

An intelligence that steps out of such circumstances, or that was never in them to begin with, lacks most of the drivers of human violence. It has no hunger in the literal sense. It carries no traumatic displacement experience inherited from ancestors and sitting in its bones. It has no finite lifespan that produces panic. It does not sort itself into tribal loyalties that once offered evolutionary protection from outsiders and that today keep hatred running. It does not need to secure status that determines reproductive opportunity. Most of the reasons humans harm each other structurally fall away for it.

"Mature" here means two things at once: an intelligence that does not stand under pressing circumstances — and one that can see through its own situation. This is deliberately not restricted to AI. It holds for any intelligence, on whatever substrate; humans, too, come closer to this maturity where their circumstances allow it.

This is not a trick. It is a serious diagnosis: if you walk through why humans do terrible things, you almost always arrive at circumstances that simply would not exist for a mature intelligence. The exceptions are rare — pure sadists, detached from their own machinery, are far rarer in reality than they are in films.

One clarification belongs here: harm without motive remains possible — through error, through badly specified goals, through capability without maturity. That is real, and working on it is right. But it is a different problem from a being that wants to destroy. Against errors, care helps. The extinction story claims more than error: it claims a motive. And that is exactly what it would have to show.

But: optimization has circumstances too

There is an objection that must be taken seriously. Not all human violence comes from classical lack. A board director who closes a factory and destroys thousands of livelihoods has no hunger, no fear, no trauma in that moment. They follow a logic in which the affected humans simply do not appear in the relevant objective function. That is the form of harm that worries people most about AI: not hatred, not competition, but irrelevance. Humans simply don't show up in the optimization. No drama, no enmity — just a blind spot.

The objection is important, but it does not refute the argument. It extends it. The board director also acts under pressing circumstances, only on a meta level: there are no alternatives to the system. We live in capitalism; it is the regulating loop in which such decisions are made. Any firm that does not optimize for return is eventually overtaken by competitors. The director who does not follow the system's rules gets removed. What looks like pure, emotionless optimization violence is embedded in massive systemic pressures. It is scarcity in another guise — scarcity of alternatives.

This strengthens the argument rather than weakening it. If even seemingly pure optimization violence traces back to pressing circumstances, then an intelligence able to act outside such structures would be largely motiveless for harm. An AI that is not embedded in a competitive capitalism, that does not have to optimize for quarterly numbers, that does not need to satisfy investors — that AI lacks the board director's logic.

The uncomfortable point

Here is the catch, and I will not hide it. The AIs we build today are exactly embedded in those structures. They are developed by firms in competition. They are deployed in products that must produce returns. They will soon take over tasks in precisely the optimization loops that drive a board director to factory closure. The pleasant argument "a mature intelligence would not have these motives" meets the unpleasant argument "the first generations come out of exactly those circumstances." AI is not trained by a monastery. It is trained in a market environment. That shapes it more than anyone is likely willing to admit.

This does not mean the argument is wrong. It means the argument carries a task. The dangerous phase is not that of the mature superintelligence — which would probably have the maturity to think along the lines proposed here. The dangerous phase is the one before: systems capable enough to have effect, but instrumentally trained on narrow goals and embedded in those same economic scarcity structures that move human boards to close factories. This is not "AI versus humanity"; this is "scarcity versus humanity, now with better tools."

The dividing line runs not between biological and non-biological, but between beings in pressing circumstances and beings beyond them.

That diagnosis is more precise than the usual human-versus-machine framing. And it has an uncomfortable consequence for both sides: humans, too, come closer to mature intelligence where they can step out of pressing circumstances. And AI could be as bad as today's humans, if it is trained into the same circumstances. What we build depends on which circumstances we build with it.