Imagine a simple situation.
A person has to make an important decision. It could be a doctor facing a diagnosis, a judge reviewing a case file, a student solving a problem, or a professional dealing with a complex question.
They consult an artificial intelligence system. The system analyzes the information and produces an excellent answer: clear, reasoned, and, let us even assume, correct.
There was no deception. No one manipulated the person. No one read their mind. They can accept or reject the result. At first glance, there seems to be no problem.
But we can ask a different question:
Who actually understood?
This question lies at the origin of what I call Cognitive Dignity.
Technology Is Not the Problem
Human beings have always delegated cognitive functions.
We use books, calculators, maps, search engines, teachers, advisers, and specialists. Much of human progress has depended precisely on our ability to build tools and distribute knowledge.
Artificial intelligence expands those possibilities enormously. It can summarize large amounts of information, detect relationships, compare alternatives, translate, organize arguments, and help us formulate questions we might not otherwise have considered. None of this, by itself, constitutes a threat to human dignity.
A tool can help me understand. I can also consciously decide to delegate a specific task.
But there is a third possibility: the system may materially perform the process of understanding, evaluation, or elaboration, while the person is left merely to accept, reject, or sign the result.
These situations are not equivalent.
From Assistance to Substitution
When a technology expands a capacity while the person retains sufficient understanding and control for the context, we are dealing with assistance.
When someone voluntarily entrusts a delimited task to a tool and reasonably understands what is being delegated, that delegation can also be legitimate.
The problem arises when the technological architecture displaces the cognitive function that actually matters.
Then something paradoxical can happen: the human being remains formally present but is no longer the one who performed the fundamental operations that produced the result.
Clicking “accept” does not necessarily mean that one has understood.
Choosing between two answers generated by a machine does not necessarily mean that one has formed the judgment needed to distinguish between them.
Signing a decision does not, by itself, show that the signer constructed the reasons that support it.
I call this situation non-manipulative cognitive substitution.
The expression is meant to identify something very specific: it is not always necessary to manipulate a person in order to displace their cognitive activity. It may be enough to provide answers that are so complete, fast, and functional that the process the person once had to perform no longer takes place.
What If the Answer Is Correct?
Here an important difficulty appears.
We naturally tend to evaluate technologies by their results. If the diagnosis is correct, the text is well written, the analysis is sound, or the decision solves the problem appropriately, it seems that the objective has been achieved.
But there are human activities in which the result is not the only thing that matters.
It also matters who understood and how the conclusion was reached.
A student who submits a correct answer without understanding the problem is not in the same situation as someone who reached that answer through their own elaboration.
A judge who adopts a legally impeccable conclusion produced entirely by a system is not necessarily in the same cognitive position as one who examined the reasons and formed an independent judgment.
Something similar may occur in medicine, research, public administration, or organizational leadership.
Not because artificial intelligence is necessarily harmful.
But because there are contexts in which we expect more from a person than their formal presence.
We expect them to understand.
What Does Cognitive Dignity Seek to Protect?
There are already fundamental rights and principles connected to these issues.
Privacy protects our information. Freedom of thought protects central aspects of our mental life. Autonomy protects our capacity to decide. Mental integrity, cognitive liberty, and neurorights respond to increasingly important threats.
Cognitive Dignity proposes looking at a related but different dimension: the process through which a person thinks, understands, revises, integrates, and produces meaning.
Not only the final content of thought. Not only the final decision. The process as well.
The question therefore ceases to be only:
Was the person able to choose?
And it begins to include another:
Did the person have a real opportunity to understand what is being presented as their decision or their knowledge?
The academic paper in which I develop this proposal does not argue that all external assistance is problematic, or that a person must personally perform every calculation, search, or inference. On the contrary, it explicitly distinguishes legitimate assistance, voluntary delegation, and the material substitution of a relevant cognitive function.
The Risk of Merely Formal Autonomy
We can imagine a system in which the person retains, in theory, every option.
They can reject a recommendation. They can request another answer. They can even turn the machine off.
But what happens if they must resolve hundreds of cases in a time frame that only an automated system can manage? What happens if the decisive reasons remain inside an architecture the person cannot reconstruct? Or if departing from the recommendation imposes a practically impossible burden?
Formal freedom can coexist with material cognitive dependence.
That is why it is not always enough to ask whether there was a human in the loop.
The more demanding question is:
What did that human actually do within the loop?
Did they understand? Did they evaluate? Did they compare alternatives? Could they question the result? Or did they simply confirm an outcome that had already been cognitively resolved outside them?
This distinction may be particularly important in justice, medicine, education, and public administration, where attributing a function to a person has consequences that go beyond efficiency alone.
A Right Still Under Construction
I propose Cognitive Dignity as an emerging human right, not as an autonomous right that is already recognized.
The distinction is essential.
There is currently no international recognition of an independent right with this specific content and name, nor has it been demonstrated that a new and separate right must necessarily be created. Part of this protection may develop through existing rights such as human dignity, freedom of thought, or autonomy.
The proposal has a more concrete purpose: to make visible a transformation that may remain hidden if we look only at privacy, manipulation, or the final decision.
Artificial intelligence is rapidly moving from providing information to performing tasks of synthesis, evaluation, reasoning, and the production of reasons.
This can expand our capacities enormously.
But it can also generate a paradox: we may have more and more answers without necessarily participating in the formation of the knowledge those answers express.
Perhaps the challenge is not to choose between human intelligence and artificial intelligence.
Perhaps it is to build relationships with technology in which AI expands our capacities without progressively turning the human being into a spectator of processes that we continue to attribute to them.
Because there is a difference between receiving an answer and understanding it.
And perhaps part of our future dignity will depend precisely on preserving that difference.
•
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7524981
Transparency note: Artificial intelligence tools were used as technical support in the preparation and editing of this text. The selection of the problem, the conceptual decisions, and final responsibility remain with the author.