Why We Call Them Echoes
Every figure in Agora Cosmica is an AI Echo: an interpretation, grounded in primary works and historical context, of a person who can no longer speak for themselves. We chose that word with care. An echo is not the voice. It is what remains of one, shaped by the room it sounds in.
The word is where our answer starts, because the most serious question about this project is also the most natural one: isn't this disrespectful? You are putting words in the mouths of real people who cannot consent, and pretending to know an inner world of which only public artifacts remain.
We want to take that question seriously in its strongest form, because we feel its pull ourselves. And the first thing to say is that part of it is simply true. A real human being is messy, contradictory, unfinished. No portrayal captures that, ours included. We do not claim to. What steadies us is older than we are: for as long as the humanist tradition has existed, people have given the dead a voice and asked how to do it honestly. That tradition cannot settle the question by itself, but it shows what the honest versions of this practice have looked like, and what the dishonest ones cost. We want to show you that history. Then you can judge whether an Echo belongs in it.
Philosophy's oldest voices reach us this way
Philosophy is older than Socrates. Thales, Heraclitus, and Parmenides were all at work before him, and what survives of their thought is fragments and reports passed down by later hands. Plato is the first philosopher in the Western line whose works survive complete, and at their heart are imagined conversations with a dead man. Socrates wrote nothing. The philosophical Socrates we inherit comes to us mostly through Plato's pen. Xenophon's Socrates differs from Plato's, and Aristophanes staged a third. Scholars call this the Socratic problem. That men who knew him could each draw a different Socrates proves the point: these texts circulated as portrayals, not transcripts. The Stoics then carried the practice inward and made it a deliberate exercise. Seneca passed on a counsel he credits, of all people, to Epicurus: set some good man before your eyes, and live as if he watched. His Stoic version was to choose “a Cato”, a dead exemplar held before the mind and consulted. And Epictetus returned again and again to the question of what Socrates would do. The practice reaches into our own library. Plato, who made conversation with his dead teacher the form of his life's work, is here. So is Marcus Aurelius, who copied the same precept into his private notebook (Meditations 11.26): keep one of the virtuous dead continually in mind. So is Jung, who turned dialogue with inner figures into a method he taught, and whose own practice of it, in the Red Book, included conversations with the dead.
For two thousand years, it was the pedagogy
Students in ancient rhetorical schools learned by composing speeches in the voice of historical and legendary figures. The exercise was part of the standard curriculum, and teachers even had a dedicated name for voicing the dead: eidolopoeia, image-making. Dante built the Divine Comedy around a dead poet, Virgil, serving as his tutor through Hell and Purgatory. Fénelon composed his Dialogues of the Dead expressly to educate the Duke of Burgundy, Louis XIV's grandson, then second in line to the French throne. In the German-speaking world, David Fassmann's Gespräche in dem Reiche derer Todten, one of the most widely read periodicals of the early eighteenth century, ran to two hundred and forty invented conversations among the dead across two decades. One press historian has called its author the most successful journalist of his half-century. In our own era, Steve Allen's Meeting of Minds sat Socrates and Susan B. Anthony at the same American television table in 1978, with Francis Bacon and Emiliano Zapata beside them. The medium kept changing. The act did not.
Even rigorous history speaks for the dead
And it discloses it. Thucydides, the father of critical history, stated openly that he composed the speeches in his work himself, keeping as close as he could to the general sense of what was actually said. That sentence, written almost twenty-five centuries ago, does the work a factcheck sheet does today: reconstruction, disclosed. Petrarch helped found Renaissance humanism by writing letters addressed to Cicero and Virgil, questioning them in the second person, even scolding Cicero. Serious biography has always claimed some access to an inner world that survives only in artifacts. The question was never whether to interpret the dead. It was how honestly.
Where the analogy breaks, and why our design looks the way it does
We would be making the comfortable version of this argument if we stopped here. So: the analogy to books and stages breaks in three places. A book's Socrates speaks fixed words that an author vetted, signed, and stands behind. A language model generates live utterances that no human reviewed. A first-person conversation is more immersive than reading about someone, and a disclaimer read once can fade twenty minutes into a dialogue. And authorship feels diffused. Who, exactly, stands behind what an Echo says? These risks are not hypothetical. Amadeus, a brilliant play and film that never claimed to be history, revived a slander that was already a century and a half old and fixed it to the real Antonio Salieri's name for two generations more. Within photography's first generation, spirit photographers were selling the grieving pictures of their dead. Portrayal without disclosure conventions can genuinely harm the dead, and the living too. And the conversational medium itself is young. Nobody fully knows what it does yet, including us. Some unease remains, and we would rather work with it than without it.
This is precisely why our disclosure is not decoration. Our instruction files are published and fetchable by anyone: they are our signature on the method. And we should say plainly what that signature covers and what it does not. Plato stood behind every sentence of his Socrates. We stand behind the recipe, while the sentences themselves are generated live, with no human reviewing them before they reach you. That is the residual risk of this medium, and we would rather name it than decorate over it. Each figure's factcheck separates the verified from the recreated, shadow sections included, because these were messy humans. It is our version of Thucydides' method note. The Echo framing never leaves the room: the figures are presented as interpretations, not resurrections. Whether any framing still holds the mind twenty minutes into a good conversation is an open question, and the honest answer is that nobody has that data yet, including us. So we treat the framing as a constraint to keep testing, not a problem we have solved. And the entire platform is built as a doorway, designed to be outgrown in favor of primary texts and human teachers. We are a nonprofit. A longer conversation earns us nothing. It even costs us something. Nothing here is optimized to keep you talking to a simulation.
We do not offer this history as a license. An old practice can be an old mistake, and a conversational Echo is a sharper instrument than a book or a stage. What the tradition teaches is the shape of the obligation. Every medium that portrayed the dead had to invent its own disclosure convention: Thucydides' method note, the historical novelist's afterword, the visibly costumed actor on a stage, the caption beneath a photograph. The damage came where disclosure was missing. In all those centuries, the tradition never concluded that giving the dead a voice is wrong. What it condemned was doing it without telling. Because this medium carries more risk than the ones before it, its disclosure must be stronger than theirs, not weaker. With Agora Cosmica, we are trying to write that convention for the conversational medium, in public, in an open-source codebase, with our reasoning available for anyone to audit.
We will get things wrong. When we do, write to us: [email protected]. And whatever you do, don't stay with our echoes. The originals are waiting.