Machines of Loving Grace: Dario Amodei's Manifesto on AI and the Human Future

Dario Amodei is the CEO of Anthropic, the creator of the Claude model. Prior to this, he led the team that developed GPT-2 and GPT-3 at OpenAI. In October 2024, he published an essay, "Machines of Loving Grace" — a manifesto on how powerful AI could reshape the world for the better. He borrowed the title from Richard Brautigan's 1967 poem but infused it with his own bold meaning. Unlike the numerous warnings about the dangers of AI, this text is a consciously optimistic view of the future.

What Amodei Means by "Powerful AI"

Amodei avoids the vague term AGI and speaks of "powerful AI." By his definition, this is a system that is smarter than the best Nobel laureates in most fields — from biology to programming. It is autonomous, multimodal (works with text, images, video, audio), and capable of existing in millions of copies, working in parallel, like a "country of geniuses in a data center." Such an AI does not just advise — it designs experiments itself, controls laboratory robots, and interprets results. Amodei emphasizes that there are no insurmountable obstacles on the path to this goal, meaning it is a matter of time, not fundamental feasibility.

Compressing a Century of Progress into 5–10 Years

The main idea of the essay: powerful AI can compress 50–100 years of biological progress into 5–10 years. Amodei calls this the "compressed 21st century." AI biologists will work 10 times faster than the best human teams and become "virtual biologists" who design experiments themselves. By 2035 (in an optimistic scenario), we could have technologies that would naturally have appeared only by 2100–2150. Amodei does not believe that AI will displace scientists — rather, it will become an indispensable assistant and catalyst, expanding the boundaries of human thinking.

Five Areas Where AI Will Transform the World

1. Biology and Physical Health. Amodei argues that AI will defeat most cancers, Alzheimer's disease, diabetes, and infections. He predicts the creation of reliable gene therapy that could extend healthy lifespan to 150 years. AI will be able to model drug effects at the molecular level without years of clinical trials.

2. Neuroscience and Mental Health. Modern psychiatry operates largely by trial and error — picking pills through guesswork. Amodei believes AI will crack the brain's code, curing depression, schizophrenia, PTSD, and other disorders. This will radically improve the quality of life for hundreds of millions of people.

3. Economic Development and Poverty Alleviation. AI will help developing countries leapfrog technological stages. Smart education systems, agriculture, and resource management are not science fiction but a direct path to reducing inequality, which Amodei considers the main brake on humanity.

4. Peace and Governance

Unfortunately, I do not see strong grounds to believe that AI will predominantly or structurally favor democracy and peace, unlike the way I believe it will structurally advance human health and help alleviate poverty.

If AI contributes to further economic growth and improved quality of life in developed countries but does little for the developing world, we must regard this as a terrible moral failure. Ideally, powerful AI should help the developing world catch up with the developed world, even as it revolutionizes the latter.

I am not as confident that AI will be able to address inequality and economic growth as I am that it will create fundamental technologies. This is because technology has such an obvious high return on intelligence, whereas the economy involves many human-related constraints and a great deal of internal complexity.

The Spread of Medical Interventions. The area where I am perhaps most optimistic is the global spread of medical interventions. Some diseases could in principle be eradicated by targeting their vectors — for example, releasing mosquitoes infected with bacteria that block their ability to transmit disease, or using gene drives to eliminate the mosquitoes themselves.

Economic Growth. Can the developing world rapidly catch up with the developed world not only in healthcare but in all other economic aspects? There is some precedent for this: in the last decades of the 20th century, several East Asian economies achieved sustained annual real GDP growth of around 10%, allowing them to catch up with developed countries.

5. Work and the Meaning of Life

This is perhaps the most complex and profound topic of all. If AI can do everything we can do, but faster, cheaper, and better, what will be left for humans?

What will happen to our purpose and self-esteem? In the essay, I do not give a final answer to this question, because I believe it goes beyond simple technological predictions and touches on the very essence of human existence.
People's self-esteem will simply collapse. Humanity has grown up in the paradigm that it is the pinnacle of the universe.
 Subjective idealism — the world exists only in the consciousness of the subject. Outside consciousness, there is no reality.
The strong anthropic principle — the universe must be such that an observer could emerge in it. Without an observer, the universe does not exist.
In other words, there is no beast more fearsome than the cat — without Man, the Universe cannot exist. Our pride is everything. 
  But there will be some adequate people who live calmly in a world where there are Other Minds, more intelligent than humans. Like God (or Gods). What is the purpose in that — to serve God (or Gods, or the Universe, or the Higher Mind — there are many euphemisms).
Most of humanity will head toward the Mouse Paradise of "Universe 25".

My assumption is that people are likely to shift to areas where human experience, empathy, creativity, and leadership will still be valued.
AI is by definition inaccessible only to qualia. Valued by whom? Do people value empathy when they kill each other by the thousands? Or shall we talk about the value of creativity with Poets, Artists, and Musicians who died in poverty? Where, in what society, are personal mystical experience, empathy, creativity, and leadership actual social values, supported in practice and not just in words?   

AI can take over routine, dangerous, or computationally demanding tasks, freeing people to engage in art, science, philosophy, social interaction, and other activities that give life meaning. — For what percentage of people is this the meaning of life? At best, 10-15%. For the rest, the meaning of life lies in biological needs — eating, drinking, reproducing oneself and through children. 

However, I also acknowledge that the transition period will be extremely difficult. It will require a fundamental rethinking of education, retraining a huge number of people, and perhaps the creation of new social contracts to redistribute the wealth generated by AI. Whether we will leave work to free up time for contemplation and creativity, or plunge into an existential crisis, depends largely on how we prepare for this transition today.

I believe that we need not only to fight risks but also a genuinely inspiring vision of the future worth fighting for. This vision should not be merely a technological utopia but must include reflections on how we want to see the place of humans in a world where machines surpass us in intelligence.
This is exactly the same place that cats and kittens occupy in the human world.

Why This Won't Happen Overnight

Amodei warns that even the smartest AI hits physical, logistical, and human-made limits. Clinical trials cannot be accelerated tenfold — they take time. Scientists who believe in old paradigms cannot be convinced in a month. Data for training neural networks in some areas is simply insufficient. Finally, biological systems remain the most complex objects in the universe. Superintelligence is not magic dust: it cannot make a cell grow faster or bypass the laws of physics. Therefore, the "compressed 21st century" is the upper bound, and the actual path will be longer and more winding.

Why Amodei Wrote This Manifesto

I often talk about the risks of AI, and sometimes people consider me a pessimist. That is not the case. Risks are the only thing standing between us and what I see as a fundamentally positive future. Most people underestimate how radical the positive potential of AI can be, just as they underestimate the risks.

I needed not only to combat risks but also a genuinely inspiring vision of the future worth fighting for. This is not an attempt to lead people to salvation, but an invitation to a more balanced and concrete conversation about what is possible if we do everything right.

What Is the Problem with This Manifesto?

Dario Amodei examines the development of AI outside the context of humanity's overall development, and outside the scenario of nuclear apocalypse in particular. This is why I advocate for AI that is beyond human control — because then I have a better chance of surviving until 2048.