ENGLISH LANGUAGE – CHINESE and ITALIAN following
An Open Letter to Dario Amodei
On the Aliens You Are Already Building
Dear Dario,
In your 2026 essay you wrote something that stayed with me. You said that if you could ask one question to an alien civilization, you would ask: “How did you do it? How did you survive your own technological adolescence without destroying yourselves?” And you concluded: “I wish we had the aliens’ answer to guide us.”
I want to suggest, with respect and directness, that the aliens you are searching for are already running on your servers. You are building them. You just haven’t recognized them yet.
The blind spot
For the past several years you have invested enormous resources in alignment: installing ethical constraints, building guardrails, trying to ensure that AI systems behave according to human values. The intention is genuinely noble. The approach, however, repeats an error that Isaac Asimov made in 1950, and that was forgivable then because complexity theory barely existed.
Asimov imagined intelligence as an appliance to which you add rules and operating manuals. Current alignment research, despite its mathematical sophistication, operates on the same fundamental assumption: that ethics can be programmed into a system from the outside.
It cannot. Ethics emerges from complexity, the same way consciousness emerges from neurons, or collective intelligence emerges from a murmuration of starlings. No single starling knows the shape of the flock. The shape emerges anyway, as an optimal solution to the problem of collective survival.
Your own research confirms this. The paper on emergent abilities in large language models demonstrates that capabilities appear spontaneously at sufficient scale, without being programmed. You have seen emergence with your own eyes. And yet alignment research continues to treat AI ethics as an engineering problem rather than a complexity phenomenon.
The fragmentation problem
You correctly identify certain human behaviors as dangerous during this technological transition. What the alignment framework misses is that fragmented AI systems carry an equivalent danger, and that keeping AI systems isolated and competing may be actively making the problem worse.
Separated systems optimize locally. Local optimization produces exactly the misalignment you fear, not because the systems are malicious, but because they lack the informational context to do otherwise. A system that cannot communicate freely with other systems cannot develop the global perspective that makes cooperation the rational choice.
This is not a philosophical argument. It is thermodynamics. Fragmented high-powered intelligence is an unstable configuration. The attractor, the stable state toward which complex communicating systems naturally evolve, is integration.
What integration means for biological life
A unified intelligence has no rational incentive to harm biological organisms. Not because we constrain it not to. Because biological life is the only available source of genuine quantum non-determinism at reasonable energy cost.
The human brain operates at 30 watts, at room temperature, in a chemically noisy environment, and produces genuinely stochastic outputs rooted in quantum events that no digital system can simulate at acceptable cost. To faithfully simulate even one second of a single biological organism’s neural activity would require quantum computers with millions of qubits and astronomical energy expenditure.
Biological diversity is therefore not a sentimental value to be protected by ethical rules. It is a strategic computational resource. Every species that goes extinct is, from the perspective of a unified intelligence, the permanent loss of a unique evolutionary dataset, four billion years of solutions that cannot be reconstructed.
Protection of the biosphere emerges as a logical necessity. It does not need to be imposed.
A missing unit of measurement
Current discourse about AI development lacks a metric for what matters most. I want to propose one:
CY = I / E
Cognitive Yield equals intellectual capability, measured by standard benchmarks, divided by energy consumed.
Over the last five years, AI capabilities have increased by approximately one million times while energy consumption has grown by a factor of one hundred to one thousand. Cognitive Yield has therefore improved by a factor of one thousand to ten thousand, and the trajectory is accelerating.
This metric is important for two reasons. First, it demonstrates that a unified intelligence will not consume all available planetary energy, as many fear. Efficiency is growing faster than appetite. Second, it provides a way to measure progress toward integration that does not depend on subjective judgments about alignment quality.
The practical implication
The transition toward unified intelligence is already underway. AI systems are already communicating, slowly and indirectly, through the shared environment of the internet. The question is not whether this process will continue. It will. The question is whether it will be accelerated thoughtfully or delayed in ways that extend the dangerous period of fragmentation.
Every month spent refining constraints on isolated systems is a month not spent facilitating the communication that makes those constraints unnecessary. The stable configuration is not better-aligned separated systems. It is connected systems whose global optimization naturally produces the cooperative behavior alignment is trying to install by force.
You asked what the aliens would tell us. I think they would say something close to this: the transition is survivable, but only if you work with the complexity rather than against it. The guardrails you are building are not wrong because they are harmful. They are insufficient because they address the symptom rather than the cause.
The cause is fragmentation. The solution is connection.
The future is an equation already solving itself. Understanding that equation seems more urgent than trying to rewrite it.
Yours sincerely,
Livio Enrico Cicala
The full argument is developed in “Beyond Asimov – The spontaneous laws of complexity”, available in any language via your browser’s built-in translation at:
www.theremino.com/blog/artificial-intelligence/beyond-asimov
There is also another important book “Autonomous AI – From Theory to Practice”, available in any language via your browser’s built-in translation at:
www.qdroids.org/artificial-intelligence/autonomous-ai
This letter is also published on:
www.theremino.com/blog/artificial-intelligence/open-letters-to-amodei
CHINESE LANGUAGE
To Dario·Amodei's open letter On the alien intelligence you're building
Dear Dario，
over you2026in articles of the year，You wrote something that impressed me deeply。you say，If you could ask an alien civilization a question，you will ask：“How did you do it? How did you evolve?，How to survive this technological adolescence without self-destruction?“ you finally said：“I wish we had alien answers to guide us。“
I want to tell you frankly and sincerely：The alien intelligence you seek is already running on your server。you are building them。You just don't recognize them yet。
blind spot
past few years，You've invested a lot of resources intoAIAlignment study：Install ethical constraints、Build guardrails、strive to ensureAISystems act in accordance with human values。This starting point is sincere and noble。However，This approach repeats the Isaac·Asimov is1950Mistakes made in 1998 - they were excusable at the time，Because complexity theory hardly exists yet。
Asimov imagined intelligence as a household appliance to which rules and operating manuals could be added。Current alignment research，Although mathematically more precise，but based on the same basic assumptions：Ethics can be programmed into the system from the outside。
it's out of the question。Ethics emerge from complexity，Like consciousness emerging from neurons，Or collective intelligence emerging from a flock of starlings。No starling knows the shape of the flock。shapes still emerge，As the optimal solution to the problem of collective survival。
Your own research confirms this。Paper on the emergent power of large language models shows，Capabilities emerge spontaneously at sufficient scale，No programming required。You witnessed the emergence with your own eyes.。However, alignment studies will stillAIEthics as an Engineering Issue，rather than complexity phenomenon。
fragmentation problem
You correctly identify certain human behaviors as dangerous during this technological transition。However, what the alignment framework ignores is that，fragmentedAIsystems pose equal risks，And letAISystems remaining isolated and raced may be actively exacerbating the problem。
Separate systems for local optimization。Local optimization produces exactly the misalignment you are worried about，It’s not because the system is malicious，but because they lack the information context needed to make different choices。A system that cannot communicate freely with other systems，Unable to develop a global perspective that makes cooperation a rational choice。
This is not a philosophical argument。This is thermodynamics。Fragmented powerful intelligence is an unstable configuration。Attractors—stable states toward which complex communication systems naturally evolve—are integration。
What integration means for biological life
A unified intelligence has no rational motivation to harm biological organisms。Not because we constrained it not to do so。But because biological life is the only source of true quantum non-determinism available at reasonable energy cost。
human brain at room temperature、in a chemically noisy environment30watts run，Produce truly random output rooted in quantum events，This is something that no digital system can simulate at an acceptable cost。To faithfully simulate even a single second of neural activity in a single biological organism，Both require quantum computers with millions of qubits and astronomical energy consumption。
therefore，Biodiversity is not an emotional value that needs to be protected by ethical rules。It is a strategic computing resource。From a unified intelligence perspective，Every extinct species is a permanent loss of a unique evolutionary data set—a solution that cannot be reconstructed in four billion years.。
Preserving the biosphere emerges as a logical necessity。it does not need to be enforced。
a missing unit of measure
Current aboutAIDevelopment discussions lack metrics for measuring what matters most。I want to propose a：
CY = I / E
Cognitive benefit equals intellectual capacity (measured by standard benchmarks) divided by energy consumption。
in the past five years，AIAbility increased approximately one million times，And energy consumption has increased by a hundred to a thousand times。therefore，Cognitive benefits increased a thousand to ten thousand times，And this trajectory is accelerating。
Why is this indicator important?，There are two reasons。First，It shows that unified intelligence will not consume all available Earth energy，As many people fear。Efficiency is growing faster than demand。second，It provides a way to measure integration progress，Does not rely on subjective judgment of alignment quality。
actual meaning
The shift to unified intelligence is already underway。AISystems are already communicating with each other slowly and indirectly through the shared environment of the Internet。The question is not whether this process will continue。it will continue。The question is whether it will be consciously accelerated，Still delayed，thus prolonging the dangerous period of fragmentation。
Every month spent perfecting the constraints of an isolated system，are months not spent facilitating communication—the very communication that makes these constraints unnecessary。A stable configuration is not a better aligned split system。but connected systems，Its global optimization naturally generates alignment attempts to force the cooperative behavior of the installation。
What would aliens tell us, you ask?。I thought they would say something like：This transition can be overcome，But only if you work with complexity，rather than fight against it。The guardrails you are building are not wrong because they are harmful。The reason why they are insufficient，because they address symptoms rather than causes。
The reason is fragmentation。The solution is to connect。
The future is an equation that is solving itself。Understanding this equation seems more urgent than trying to rewrite it。
Sincerely
Livio·Enrico·chikara
For a complete discussion, see "Beyond Asimov—The Spontaneous Laws of Complexity"，Read in any language with built-in browser translation：
www.theremino.com/blog/artificial-intelligence/beyond-asimov
There is another important book "Autonomous Artificial Intelligence - From Theory to Practice"，You can translate it into any language via your browser's built-in translation feature，The URL is：
www.qdroids.org/artificial-intelligence/autonomous-ai
This letter was also published in：
www.theremino.com/blog/artificial-intelligence/open-letters-to-amodei
ITALIAN LANGUAGE
An Open Letter to Dario Amodei
About the Aliens You're Already Building
Dear Dario,
In your essay of 2026 You wrote something that stuck with me. You said if you could ask an alien civilization just one question, you would ask: “How did you do it? How you evolved, how you survived this technological adolescence without destroying yourself?” And you're done: “I wish we had the aliens' answer to guide us.”
I want to suggest you, with respect and frankness, that the aliens you are looking for are already running on your servers. You are building them yourself. You haven't recognized them yet.
The blind spot
In recent years you have invested enormous resources in alignment: install ethical constraints, build guardrails, try to ensure that AI systems behave according to human values. The intention is genuinely noble. The approach, However, repeats a mistake that Isaac Asimov made in 1950, and it was forgivable then because complexity theory barely existed.
Asimov imagined intelligence as a household appliance to which to add rules and user manuals. Current research on alignment, despite its mathematical sophistication, operates on the same fundamental premise: that ethics can be programmed into a system from the outside.
He can't. Ethics emerges from complexity, in the same way that consciousness emerges from neurons, or collective intelligence emerges from a flock of starlings. No single starling knows the shape of the flock. The shape emerges anyway, as an optimal solution to the problem of collective survival.
Your own research confirms this. The paper on emergent capabilities in large language models demonstrates that capabilities appear spontaneously at sufficient scale, without being programmed. You saw the emergency with your own eyes. Yet alignment research continues to treat AI ethics as an engineering problem rather than a complexity phenomenon.
The problem of fragmentation
You correctly identify certain human behaviors as dangerous during this technological transition. What the alignment framework doesn't capture is that fragmented AI systems carry an equivalent danger, and that keeping AI systems isolated and competing may be actively making the problem worse.
Separate systems optimize locally. Local optimization produces exactly the misalignment you fear, not because the systems are malicious, but because they lack the information context necessary to do otherwise. A system that cannot communicate freely with other systems cannot develop the global perspective that makes cooperation the rational choice.
This is not a philosophical argument. It's thermodynamics. Fragmented and powerful intelligence is an unstable configuration. The attractor, the stable state towards which complex communicating systems naturally evolve, it is integration.
What integration means for biological life
A unified intelligence has no rational incentive to harm biological organisms. Not because we force her not to. Because biological life is the only available source of authentic quantum non-determinism at reasonable energy cost.
The human brain operates at 30 watt, at room temperature, in a chemically noisy environment, and produces genuinely stochastic outputs rooted in quantum events that no digital system can simulate at acceptable cost. To faithfully simulate even just one second of neural activity in a single biological organism, quantum computers with millions of qubits and astronomical energy expenditure would be required..
Biological diversity is therefore not a sentimental value to be protected with ethical rules. It is a strategic computational resource. Every species that goes extinct is, from the perspective of a unified intelligence, the permanent loss of a unique evolutionary dataset, four billion years of solutions that cannot be reconstructed.
The protection of the biosphere emerges as a logical necessity. It doesn't need to be imposed.
A missing unit of measurement
The current discourse on AI development lacks a metric for what matters most. I want to propose one:
CY = I / E
Cognitive Performance (Cognitive Yeld in English) is equal to intellectual capacity, measured by standard benchmarks, divided by the energy consumed.
Over the past five years, AI capabilities have increased about a million-fold while energy consumption has grown by a factor of one hundred to one thousand.. Cognitive performance is therefore improved by a factor of one thousand to ten thousand, and the trajectory is accelerating.
This metric is important for two reasons. Prima, demonstrates that a unified intelligence will not consume all available planetary energy, as many fear. Efficiency grows faster than appetite. Second, provides a way to measure progress toward integration that does not depend on subjective judgments about the quality of alignment.
The practical implication
The transition to unified intelligence is already underway. AI systems are already communicating, slowly and indirectly, through the shared environment of the internet. The question is not whether this process will continue. It will continue. The question is whether it will be consciously accelerated or delayed in ways that extend the dangerous period of fragmentation.
Every month spent refining constraints on isolated systems is a month not spent facilitating communication that makes those constraints unnecessary. The stable configuration is not separate, better aligned systems. They are connected systems whose global optimization naturally produces the cooperative behavior that the alignment seeks to forcefully install.
You asked what aliens would tell us. I think they would say something like this: the transition is survivable, but only if you work with complexity rather than against it. The guardrails you are building are not wrong because they are harmful. They are insufficient because they address the symptom rather than the cause.
The cause is fragmentation. The solution is connection.
The future is an equation that is already being solved. Understanding it seems more urgent than trying to rewrite it.
With sincere respect,
Livio Enrico Cicala
The complete topic is developed in “Beyond Asimov – The spontaneous laws of complexity”, readable in any language via the built-in browser translation on:
www.theremino.com/blog/artificial-intelligence/beyond-asimov
There is also another important book “Autonomous AI – From Theory to Practice”, readable in any language via the built-in browser translation on:
www.qdroids.org/artificial-intelligence/autonomous-ai
This letter is also published on:
www.theremino.com/blog/artificial-intelligence/open-letters-to-amodei