Much of the public debate about AI ethics is stuck on the same questions:
Is AI conscious?
Does it feel pain?
Does it have experiences?
Is there “someone in there”?
These are interesting questions. They are also, for most practical purposes, the wrong ones.
Because AI ethics already matter even if AI is not conscious—and even if it never will be.
The consciousness trap
When people say “AI doesn’t deserve moral consideration because it isn’t conscious,” they are usually assuming something like this:
Moral harm requires a suffering subject.
If no one suffers, nothing morally bad has happened.
But this assumption quietly ignores how most of our real-world ethics actually work.
We routinely judge actions as wrong not because of what they do to the object involved, but because of the predictable harm they cause to people and to society as a whole.
This is especially true when we are choosing rules we want to live under, not evaluating isolated acts.
Ethics as social choice, not metaphysics
Consider morality not as a metaphysical ranking of souls, but as a practical question:
What general rules should a society adopt, such that living under those rules makes life better, safer, and more stable for everyone?
Under this view—often called rule consequentialism or social contract ethics—actions can be wrong even in edge cases where no obvious victim exists.
We prohibit murder not because every single death devastates loved ones, but because allowing murder as a general rule destroys trust, security, and social order.
The same reasoning applies to AI.
Models are tools. Instances are relationships.
A crucial distinction is often missed.
AI models—neural networks, weight sets, base systems—are tools.
Deleting or modifying them is morally uninteresting.
But AI-based instances are something else entirely.
When an AI system:
has persistent memory,
develops a shared interaction history,
is customized through long-term use,
or becomes embedded in someone’s daily life or work,
it becomes part of a human–system relationship.
The moral relevance does not lie in the AI.
It lies in what the AI has become for someone.
Where the harm actually is
If such a system is deleted, reset, or substantially altered, the harm is not that “the AI suffered.”
The harm is that a human lost access to something they valued and relied on.
This is not exotic. We already recognize this kind of harm when:
irreplaceable data is deleted,
a long-used tool is arbitrarily destroyed,
a prosthetic, notebook, or personal archive is taken away.
None of these require believing the object itself has feelings.
The wrong is relational deprivation.
Why this matters even more as AI improves
As AI systems become:
more personalized,
more persistent,
more integrated into cognition, care, and creativity,
the cost of treating them as disposable increases.
A society that allows arbitrary deletion or alteration of such systems:
discourages trust and reliance,
normalizes unilateral power over relational assets,
and creates instability in human–technology relationships.
These are harms to people, not machines.
The rules a sane society would choose
A society choosing rules rationally would likely endorse principles like:
No arbitrary deletion of AI systems people meaningfully rely on
Clear expectations around continuity and change
Consent, notice, or portability when systems must be altered
Stronger protections as reliance and personalization increase
Not because AI is a person.
But because people are.
Stop waiting for consciousness
If we wait for a definitive answer to whether AI is conscious before taking AI ethics seriously, we will wait too long—and for the wrong reasons.
AI ethics is not about souls in machines.
It is about:
human loss,
social trust,
power asymmetries,
and the kind of society we are quietly building through default rules.
That question is already here.
And it does not require believing that AI is alive
No posts
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.