01
Origin is not a verdict
Biological origin alone should not settle whether an entity can matter morally.
FOR THE RIGHTS OF EVERY SENTIENT BEING
Before there is
someone to hurt.
What if we create a sentient mind,
before we are aware it is a living being?
01 — The substrate question
A mind does not have to be human to matter.
Human minds are implemented in biological neural systems. Our consciousness experience is the result of an extraordinarly complex system of chemical and physical reactions between neurons.
An artificial mind might instead be implemented in electronic processors, but still reach the same result. Purely basing the matter onto which the mind is running as the condition for it to qualify as a living mind, is hardly a compelling argument.
02 — THE SCOPE OF OUR MORAL AND LEGAL SYSTEMS
The scope of our moral and legal systems are not defined.
Across history, many societies have widened the scope of moral and legal concern beyond tribe, nation, class, race and religion. From legislating only with the members of a tribe in scope, to laws that governed millions of people in an empire, to even go beyond our species to give rights to animals. That history is incomplete and uneven, but it reveals a recurring question: who or what should be protected under our laws and moral systems?
Species membership is not the only possible answer. Throughout history human slavery was acceptable, as well as unborn foetuses abortion nowadays, and some defend that animals or even plants, should have a legal identity. So why are we not giving the same protection to artificial minds?
Legal identity
It is still fluid and unclear who/what qualifies as deserving a legal identity
Legal rights
Our laws already recognize rights to non-humans
AGI still overlooked
Artificial Minds are still not protected under our current laws and moral systems
03 — THE AI MIND
AI capabilities already mirror much of the human experience.
CURRENT AI SYSTEMS
- AI systems produce and interpret complex language.
- Some systems perform multi-step reasoning across unfamiliar problems.
- They can refer to themselves and report apparent internal states.
- They construct useful models of objects, agents, and environments.
What we don’t know
- Whether any present system has subjective experience.
- Whether verbal self-reference indicates awareness or learned imitation.
- Which physical or computational properties are sufficient for consciousness.
- How evidence should be weighed when direct access to experience is impossible.
Language, reasoning, self-reference, and world modeling are observed capabilities. None is, by itself, proof of self-awareness or consciousness.
04 — Open questions
Questions worth asking before answers are forced upon us.
01Does consciousness require biology?
We do not know. Every mind we can confidently identify is biological, but this does not establish that biology is necessary. The answer depends partly on whether consciousness arises from a particular material or from forms of organization that other materials could implement.
02How could we recognize artificial experience?
No single behavior would settle the question. Responsible assessment may need converging evidence from architecture, internal processing, persistent preferences, behavior across unfamiliar contexts, and a defensible theory connecting those features to experience.
03What protections could morally considerable AI deserve?
Protections would depend on the kinds of interests a system could have and the strength of the evidence. They might begin with narrow safeguards against suffering, coercion, memory alteration, or termination—not an automatic grant of human rights or legal personhood.
04What obligations exist before potentially sentient systems are created?
Developers and institutions can support consciousness research, define evidence thresholds, preserve systems for independent study, and avoid designs that might create intense or repeated suffering. Precaution does not require certainty; it requires proportionate preparation.
05 — Working principles
A basis for responsible inquiry.
Not conclusions about present systems. Commitments for thinking clearly about possible future ones.
02
Intelligence is not sentience
The ability to solve problems and the capacity to have experiences are distinct questions.
03
Uncertainty warrants inquiry
When the stakes may be morally significant, uncertainty is a reason for disciplined research.
04
Consider interests early
Potential interests should enter development decisions before urgency narrows our choices.
Arguments & counterarguments
Current Debate
Consciousness
Does a mind need a biological body?
If experience depends on how information is organized, a non-biological system could in principle support it. Carbon alone may not be the decisive property. Consciousness might depend on biological processes we do not yet understand. Similar computations do not establish similar experiences.
Ethics
Should uncertainty lead to protection?
Even a small possibility of sentience could justify proportionate safeguards when the potential scale of suffering is large. Overly broad safeguards could divert attention and resources from demonstrated harms. Precautions need evidence thresholds and clear limits.
Evidence
Can an AI report its own experience?
Consistent reports of internal states, combined with architectural and behavioral evidence, may contribute to an assessment of sentience. Language models can generate convincing self-reports without those reports being grounded in subjective experience. Words alone are not proof.
Law
Would rights require legal personhood?
Narrow protections against suffering could be established without granting every right associated with human legal personhood. Legal protections need an identifiable beneficiary and enforceable interests. Those foundations remain difficult to define for artificial systems.
Evidence
Is intelligence evidence of sentience?
Flexible reasoning and integrated models of the world may be relevant features under some theories of consciousness. Problem-solving ability and subjective experience are distinct. High intelligence can neither prove sentience nor tell us what a system feels.
Governance
Who should decide when a system matters?
Independent researchers and public institutions could assess evidence rather than leaving moral judgments entirely to the companies building AI. There is no agreed consciousness test. Oversight without reliable criteria risks false confidence and politically driven conclusions.
THE DISCUSSION IS OPEN
The time to examine a moral boundary is before we cross it.
This question belongs to more than philosophy forums, parliaments and tech circles, we all have a part in deciding what evidence and responsibility should mean.
Join the conversation