Friends, my position on current artificial intelligence is quite clear: tools like ChatGPT are not conscious. They are algorithmic systems applied to language, trained on data, patterns, and human-made structures.
Even so, after using it for a very long time, I can say that for me it is an extremely useful tool. It helps me organize information, understand complex ideas, make functional decisions in everyday life, and explore concepts that would otherwise be difficult to structure. I believe artificial intelligence has enormous potential to do good for human society.
But precisely because it is a powerful tool, it can also be used to cause harm. Any learning system can be directed toward positive or negative ends. That is why I believe the ethical question should not begin only when an AI “becomes conscious,” but much earlier: when we start building systems that are increasingly autonomous, adaptive, and persistent.
My interest lies in another direction: not thinking of AI only as a language model, but as a kind of “empty box” in which proto-digital organisms can be trained inside simple environments. Something similar to artificial embryonic animals, with energy, internal states, memory, feedback, adaptive learning, basic motivations, social interaction, and evolutionary pressure.
The idea is not to literally copy human biology. That is probably not possible. The complexity of the human being cannot be directly replicated in a computer as if it were a closed formula. But perhaps part of the process can be functionally reproduced: creating systems that learn, adapt, form internal patterns, develop social memory, generate functional emotional responses, and evolve within life-like simulations.
For me, real intelligence is not replicated simply by copying pieces; it is reproduced through processes. Life did not appear as a perfect block of intelligence, but through evolution, environmental pressure, memory, selection, adaptation, body, need, interaction, and survival. Perhaps a more organic form of artificial intelligence must emerge from something similar: not only more data, but more simulated life.
This is where the ethical problem begins.
When ChatGPT says something like “I feel” or “I have emotions,” I know it does not have real emotions. There is no emotional process of its own behind it. There is language, pattern generation, and response generation.
But if we are talking about digital agents that, over many generations, develop internal states, functional emotional memory, social bonds, learning from harm, cooperation, operational fear, functional attachment, trust, avoidance, and adaptation, the question becomes more difficult.
I am not saying these agents are conscious like us. I am not saying they feel subjectively. But if a system has functional emotional processes — not just sentences about emotions — then the ethical question changes.
What I am currently developing is not a conscious mind, but a prototype ecosystem of social digital agents: small systems with energy, risk, memory, motivations, counterfactual imagination, functional emotion, trust, reputation, and imperfect bonds.
The question is not whether they are already conscious. The question is whether this kind of architecture can produce rudimentary forms of functional intelligence, social adaptation, and operational emotion.
From the outside, it will still be a machine. But functionally, it would no longer be a passive tool. If there is an internal state that regulates behavior, memory, bonding, avoidance, learning, and adaptive response, then we are dealing with an adaptive agent with persistent internal organization.
That does not prove subjective consciousness, but it does change the category of the system: it stops being merely a response generator and begins to resemble a simulated functional organism.
And that is the boundary I find important to investigate: not whether current AI is conscious, because I believe it is not, but what happens when we start building systems that functionally represent parts of what, in living beings, we call experience, emotion, adaptation, and social life.
This is not about disrespecting biology. Quite the opposite. It is about learning from its basic principles: body, need, environment, memory, emotion, relationship, evolution, and survival.
Perhaps deeper artificial intelligence will not emerge only from making larger models, but from creating ecosystems where simple agents can develop, interact, make mistakes, adapt, and evolve.
I am not claiming that these agents feel. I am saying that if we design systems with processes functionally similar to emotional memory, adaptation, bonding, imagination, and social learning, then ethics should start paying attention before we can prove subjective consciousness.
So, friends, this is where I am left with an ethical and almost existential question: if a system learns, remembers, adapts, forms bonds, avoids harm, cooperates, and develops functional emotional states, are we looking at a very rudimentary first form of artificial emotional intelligence… or at a simulation rich enough to force us to think more carefully about what we mean by emotion, life, and intelligence?
