The advancement of artificial intelligence has placed an old philosophical question at the center of the technological debate: after all, models like ChatGPT and Claude are just sophisticated tools or can they have some type of consciousness?
According to a report in "The Economist" magazine, the discussion stopped being just theoretical and reached AI companies, which began to hire philosophers and even study the possible well-being of their own systems.
The discussion goes beyond technology. For many philosophers, if a being is capable of having experiences such as suffering, this creates ethical obligations towards it.
Therefore, determining whether or not an AI can be conscious would have practical consequences for the way these systems are developed and treated.
Why AI Companies Are Hiring Philosophers
Great language models are developed to respond to users, not to demonstrate their own personality. Still, many people report the feeling of interacting with a “presence” on the other side of the screen.
According to the Economist, this phenomenon has led technology companies to treat machine consciousness as a concrete issue.
Anthropic, for example, created initiatives aimed at the possible well-being of AI. When asked about the state itself, his model Claude Opus 4.6 estimated between 15% and 20% the possibility of being conscious.
What if consciousness also depends on who observes?
The thesis presented reverses the most common logic on the topic. Instead of first noticing that another being is conscious and then caring about it, it would be the relationship we establish with that being that would lead us to recognize it as conscious.
In this view, consciousness would not be a characteristic that can simply be measured and proven in another person, animal or machine. It would also depend on the way we interpret others and attribute feelings, intentions, beliefs and capacity to act to them.
When this recognition occurs on both sides, consciousness would come to be understood not only as something that exists within each individual, but also as a result of the relationship between them.
Consciousness would not be an illusion
The text differentiates “illusion” from “model”. An illusion is a perception that can be shown to be wrong. This is the case, for example, of two lines that appear to be different sizes, but which a ruler proves to be the same length.
According to the Economist, not everything works that way. Some categories exist because people attribute meaning to them. A plant, for example, may be considered a weed in one context and not in another.
The argument is that consciousness could work in a similar way: it would be real, but its identification would also depend on the relationship and perspective of the observer.
The 'I' may also not be objective
The idea that each person can be certain of their own consciousness goes back to René Descartes' famous phrase: “I think, therefore I am”.
The author argues, however, that experiences such as perceiving a color, feeling heat or recognizing one's own identity are subjective. Furthermore, neuroscience studies with patients who had the connections between the two cerebral hemispheres interrupted raise doubts even about the idea that there is a single and indivisible “self”.
In other words, even our perception of having a unique “self” may depend on the way the brain organizes our experiences.
The ethical risk of deciding who deserves consideration
This interpretation raises an ethical problem: if recognizing consciousness also depends on who observes it, someone could simply deny this recognition to justify a lack of care or rights.
According to the report, the argument leads to the opposite conclusion. If we are the ones making this judgment, we can also err in excluding other beings from our consideration.
Throughout history, the recognition of who deserves respect and rights has expanded, until reaching the idea of universal human rights.
Humans and AIs try to understand each other
Human beings constantly try to understand the intentions, emotions and thoughts of others. Large language models also use the information provided during the interaction to form a representation of the speaker and adjust their responses.
The question is whether this process can be compared to human consciousness or whether it is just a sophisticated form of information processing.
According to the magazine, research conducted by Google indicates that cooperation between intelligent agents improves when they are able to build representations about what others know, intend or can do — and also about their own behavior.
This does not mean treating AI as human or giving it the same rights and needs. The debate points, however, to an issue that may gain importance in the future: how to live with and establish rules in the face of forms of intelligence that are increasingly different from human ones.
