“In the Schillerian sense, I am absolutely an optimist. The optimist does not believe everything will turn out well. He knows that not everything can go wrong.” With this sentence, Gwendolin Kirchhoff, in her debate with Joscha Bach (Everlast AI, 2026), rejected one of his terms without ever naming it. Bach had said earlier: “Personally, I see more reason for hope.” Between those two sentences lies the difference between optimism and over-optimism. The first reckons with the world as it is — mortal, uncertain, shot through with tendencies that do not all point in the same direction. The second extrapolates a salvation expectancy from a technical trend and calls it realism.
#The temptation to think like an engineer where being is at stake
Over-optimism is not the expectation that computers will get faster. That expectation is regularly correct; it follows the public roadmap of the semiconductor industry. Over-optimism begins where the same logic gets applied to domains in which it holds no ontological jurisdiction: to consciousness, to mortality, to the meaning of a human life. To say that a computer will compute ten times faster in ten years is to make a technical claim. To say that a computer will be conscious in ten years is no longer a technical claim. The speaker has left the scale on which his optimism was checkable and crossed into another domain without naming the rules of translation between the two.
That is the signature of over-optimism: an extrapolation that exceeds its own range of validity without marking the transgression. Ray Kurzweil is the canonical figure of this style of thought. His prediction of biological immortality by 2032 (cf. Kurzweil, 2005; 2024) has no empirical basis comparable to the basis of his processor forecasts. Yet it is presented in the same breath, with the same table, on the same exponential curve. The listener carries the trust earned by the first claim and silently transfers it to the second. That transfer is not an inference. It is an act of faith.
#Optimism and over-optimism
In 1795, in his Letters on the Aesthetic Education of Man, Schiller wrote that the human being is suspended between sensuous drive and formal drive, and that human freedom is realised only in the play-drive — in the aesthetic mediation of the two forces (cf. Schiller, 1795). The Schillerian optimist does not hope for the good. He knows that the world carries within itself the capacity for the good, and that this capacity unfolds only where the human being takes up his own task. Optimism in this sense is a form of responsibility. It does not say: “The machine will fix it.” It says: “It is up to you, up to me, up to us.”
Over-optimism is the outsourcing of that responsibility to a technical bearer. What theology once knew as expectancy of redemption, what the Enlightenment understood as the pedagogical project of self-formation, gets here transferred onto a machine that is meant to arrive at some later date. The structure is religious: salvation expectancy, a date, the promise of overcoming death. Only the bearer has changed. What the Theosphäre once delivered, the Technosphäre is now meant to redeem. The hope migrating here has, in its depth, remained the same.
The difference between Schiller’s optimism and over-optimism does not lie in the magnitude of hope but in its addressee. Optimism reckons with the living — with people who act, with relationships that hold, with a reality that opens up in the doing. Over-optimism reckons without the living. It reckons with data structures, with scaling laws, with hypothetical substrates such as Computronium, in which every atom is meant to serve as a logic gate. Whoever calculates this way has already removed the living from the equation before writing the first line.
#Spengler and the Promethean impulse
In 1931, in Man and Technics, Oswald Spengler described the mechanism of this outsourcing with precision. Modern man, he wrote, is “too shallow and too cowardly” to bear the fact of the transience of all that lives, and so hides behind ideals in order to see nothing (cf. Spengler, 1931). Spengler is not aiming at pessimism but at structural analysis: hope becomes fanatical wherever it is meant to cover over a reality that appears unbearable. Kurzweil hides behind exponential curves rather than ideals. The form has modernised; the psychic operation is the same.
In 1967, in The Myth of the Machine, Lewis Mumford showed that the mega-machine does not begin with the industrial revolution but already in the early high cultures, as a form of organising human labour that reduces the individual to a function (cf. Mumford, 1967). What he describes is the willingness to understand oneself and others as cogs in a machine, because the machine promises something the single life cannot hold: immortality of the deed, permanence of order, the overcoming of finitude. Over-optimism is the mood in which that willingness reaches its peak. It does not regard itself as willingness; it regards itself as sober prognosis.
In conversation with Joscha Bach, Gwendolin named this nexus the “Promethean impulse”: the impulse to destroy and replace the living, or to destroy it accidentally in the attempt to replace it. What presents itself as technical hope is, on this reading, a form of enmity against life that does not recognise itself, because it takes itself for the progressive healing of what it is destroying.
#The diagnostic: localising rather than refuting
Over-optimism cannot be refuted by better data. Whoever holds a prediction that swaps out its mechanism but keeps its date will not be deflected by data after the next mechanism swap either. In The Singularity Is Near (2005), nanotechnology was supposed to bring biological immortality. In The Singularity Is Nearer (2024), nanotechnology is no longer mentioned; now AI-driven drug development is supposed to redeem the date 2032. The mechanism has changed completely. The date has stayed the same. That is how to recognise the difference between a prognosis oriented toward the world and a prophecy defending its appointment.
Stuart Armstrong and Kaj Sotala showed in 2014 systematic over-optimism patterns in AI predictions; independent assessments of Kurzweil’s 105 predictions for 2019 found hit rates between 7 and 24 per cent, not the 86 per cent he claimed for himself. The moment a prediction depended on biology, physics, or human behaviour rather than on transistor scaling, he was wrong. The decisive point is not the rate but the pattern: where the over-optimist is certain, he has changed domain. Certainty earned in one realm he carries into another in which it does not hold.
The philosophical task is not to refute every individual prediction. It is to localise the shift at which a technical claim tips into a metaphysical one and to make visible that this tipping point exists at all. That work belongs to contextual disclosure: to the uncovering of the unconscious presuppositions a discourse operates on. Over-optimism operates on the presupposition that there are no ontologically distinct domains, that everything is in the end a question of sufficient computational power. That presupposition is not an observation but a metaphysical positing.
#The form of hope
A civilisation that can no longer distinguish between warranted expectation and salvation expectancy has lost its power of judgement — not in the sense of formal logic but in the sense of that capacity called judgement: the ability to bring a case under the concept that fits it. Whoever discusses consciousness as computation has placed the case under the wrong concept before the conversation begins. The discussion then proceeds without result, because its object was already lost in the choice of language.
Schiller’s optimism preserves the distinction between what will be and what ought to be. He knows that not everything can go wrong because he knows that forces are at work in the living that are not at the disposal of the subject. This form of hope is not naive. It is held — held by a world the human being did not create and which carries him without his doing. Whoever is optimistic in this sense does not need a singularity. He does not need a date at which salvation arrives. He already stands on the ground of what natural philosophy calls the living, and works onward from there, in the span a single life affords.
Once you have seen the difference between optimism and over-optimism, you see it everywhere: in the language of AI investment, in the dates of the longevity industry, in the self-assurance of the futurists. The point is not to make hope smaller, but to give it back its addressee. The world is not healed by machines. It is healed, if at all, by what human beings do, with one another, in the awareness of their own finitude. Related entries on pathogenesis-not-progress, technological singularity, and Longevity Escape Velocity deepen the individual symptoms of this confusion.