10/04/2026
An excerpt from latest article linked in comments.
How do we treat something that cannot tell us no?
Consent is permission, within a relationship, to act. At its core, consent places a limit on power. It says: Just because I can do something does not mean I am entitled to do it.
When two parties possess relatively equal power, those limits can be negotiated. But the greater the imbalance of power, the greater the responsibility falls upon the powerful to restrain themselves.
The question is larger than whether the river should have a voice. Or whether young women should have a voice.
What does power owe to that which it has the ability to overpower?
We are beginning to confront a strange inversion of that question as we build increasingly capable artificial intelligence. AI researchers use the term CORRIGIBILITY to describe a quality we desperately hope advanced systems will retain: a willingness to accept correction, redirection or shutdown by humans even when doing so interferes with the objective the system is pursuing.
Think about what we are asking.
We want something more powerful than ourselves to recognize our boundaries. We want it to stop when we say no. We want our vulnerability to create an obligation in the powerful rather than an opportunity.
It is difficult not to turn that question back on ourselves. When nature resists us, are we corrigible? When a river floods the boundaries we constructed for her, do we reconsider where we built, or simply build a higher levee? When an ecosystem cannot sustain what we are extracting from it, do we change our objective? When another human being says this is my body, my boundary, my life, do we understand that our ability to exert power does not give us the natural right to do so?
Before asking whether artificial intelligence will respect the limits human beings place upon its power, perhaps we should examine how faithfully we have respected the limits placed upon ours.
