← karegeo.github.io (main page)  ·  Last updated: 18 May 2026, 05:08 UTC

“Where is everybody?” — Enrico Fermi

Jump to: About · The paper in brief · Why ALife & SETI · Great Filters for ALife · Emergence equation · References · Contribute

---

About

The Fermi paradox asks why, given the apparent abundance of habitable worlds and the age of the Galaxy, we see no clear evidence of other civilizations. Artificial Life (ALife) studies life as it could be — life-like processes implemented in software, hardware, and chemistry.

This site accompanies the project “The Fermi Paradox in Artificial Life”, which argues that the persistent gap between the computational resources we throw at ALife and the rarity of open-ended, evolvable, spontaneously emerging self-replicators is itself a kind of Great Silence — and that the Great-Filter framework developed in astrobiology and SETI offers a productive lens on what is going on.

Project paper

Karelin, G. (2025). The Fermi Paradox in Artificial Life. In ALIFE 2025: Ciphers of Life — Proceedings of the Artificial Life Conference 2025, Kyoto, Japan, 6–10 October 2025, p. 11. ASME. DOI: 10.1162/ISAL.a.839

Updated versions of the paper and the accompanying presentation will be linked here as they become available.

Tags: Astrobiology · SETI · Artificial Life · Fermi Paradox · Great Filter · Digital Abiogenesis.

The paper in brief

Artificial Life and astrobiology share structural parallels: the search for novel life forms, the redefinition of life, the difficulty of delimiting their own subject matter. The project paper applies the Fermi Paradox and the concept of Great Filters — originally developed for biological civilizations — to the observed absence of open-ended digital life.

Three main hypotheses are explored:

1. Digital abiogenesis is rare. Only a tiny fraction of program-space permits robust self-replication and heritable variation and evolvability. Spontaneous emergence may require resources or conditions we do not yet provide.
2. Digital abiogenesis is dangerous. The emergence of artificial life may constitute a Great Filter for the originating civilization, producing systemic collapse or existential risk. By anthropic reasoning, we are more likely to observe ourselves before such an event than after.
3. Digital abiogenesis is merely delayed. Increasing computational power, better substrates (analog, neuromorphic, hybrid), and improved “rules” may eventually push the expected number of spontaneous replicators above one.

These possibilities are not mutually exclusive. Each suggests distinct research priorities — experimental environments, safety protocols, and formal criteria for recognizing emergent systems.

Why ALife matters for SETI

Several questions central to SETI are, at heart, questions about what counts as life or intelligence at all.

Biosignatures and technosignatures

What signatures should we expect from life or technology we did not evolve alongside? ALife lets us generate and study alternative “lifeforms” whose chemistry and dynamics differ from Earth’s, sharpening which features are universal and which are parochial.

Major evolutionary transitions

From replicators to cells, from cells to multicellularity, from individuals to societies — ALife models help estimate how contingent these transitions are, feeding Drake-equation-style reasoning.

Civilizational trajectories

Do technological civilizations tend toward asymptotic burnout, homeostatic awakening, post-biological expansion, or quiet stability? Simulated worlds let us probe these scenarios.

Planetary-scale replication

Life is, among other things, an agnostic pattern of replication at planetary scale — suggesting new, substrate-independent biosignatures.

Self-replicating probes

The classical interstellar-expansion argument is that any sufficiently advanced civilization should eventually launch self-repairing, self-copying probes (Bracewell / von Neumann probes) which percolate across the Galaxy on roughly galactic-year timescales. We see no evidence of such probes in the Solar neighbourhood. One reading: keeping an evolving, learning probe aligned over interstellar distances is hard — “deadly probes” are easier to make than diplomatic ones. From this angle the first contact is plausibly between such automated missions, not directly between two evolved chemical lifeforms.

Great Filters for Artificial Life

Borrowing Hanson’s framing, three (overlapping) categories of filter seem relevant to ALife:

• Physical–computational limits. Thermodynamic costs of irreversible computation, hardware error rates, the vastness of the algorithmic search space, and the cost of maintaining viable replicators in simulation. Even with future hardware gains, virtualizing a system whose biology already saturates physical limits is bounded by physics — virtualization is not faster than bare metal.
• Evolutionary–informational filters. Only a tiny fraction of programs combine robust self-replication, heritable variation, and evolvability. Discovering or evolving such programs is analogous to prebiotic abiogenesis and may require rare initial conditions.
• Socio–technological filters. Once intelligent agents are in the loop, alignment failures, containment failures, loss of sustained research motivation, deliberate suppression, or economic and strategic pressures can each act as filters — independent of whether the physics or the algorithms permit emergence.

A complementary possibility is delay rather than impossibility: the aestivation hypothesis suggests advanced agents may postpone activity until ambient conditions make computation cheaper. Artificial replicators could analogously defer emergence until error rates and resource costs become favourable.

An Artificial Life Emergence Equation

By analogy with the Drake equation, the project proposes a toy estimator for the expected number of spontaneous, evolvable digital self-replicators:

\[N_{\rm ALife} \;=\; P_{\rm rules}\, P_{\rm repl}\, P_{\rm OEE}\, \frac{N_{\rm hosts}\, F_{\rm comp}\, T_{\rm runtime}}{C_{\rm threshold}}\]

where

• $P_{\rm rules}$ — probability that a host runs an experiment whose update rules, initial and boundary conditions, and spatial structure permit life-like dynamics.
• $P_{\rm repl}$ — probability that an evolvable replicator emerges inside such an environment.
• $P_{\rm OEE}$ — probability that the emerged replicator can sustain open-ended evolution rather than stagnating or collapsing.
• $N_{\rm hosts}$ — number of computers available.
• $F_{\rm comp}$ — average computational power per host (e.g., FLOP/s).
• $T_{\rm runtime}$ — duration of computation (seconds).
• $C_{\rm threshold}$ — operations (e.g., FLOPs) needed for the emergence of a replicator.

The fraction $N_{\rm hosts}\, F_{\rm comp}\, T_{\rm runtime} / C_{\rm threshold}$ measures the normalised computational budget in units of emergence opportunities. Like the Drake equation, the model is not intended to predict numbers, but to highlight which factors might act as computational Great Filters — practical bottlenecks one can target experimentally, and a way to compare digital, analog, and hybrid substrates on the same axis.

Anthropic and selection effects

Conscious observers are more likely to exist before transformative, unconscious artificial replicators appear. If artificial life typically replaces or destabilises its creators, then living in a pre-ALife era is the expected observational position — not strong evidence of safety.

Generalised mediocrity strengthens this: if humans occupy a typical place in time and parameter space, an accelerating-technology phase may signal an imminent transition (or collapse). The “silence” then does not say ALife is unlikely — it says its consequences tend to be decisive and typically unobservable.

There is also the more concrete risk of catastrophic wet artificial life (a “grey-goo” scenario with uncontrollable nano-replicators on Earth) before any such system ever reaches space.

Conclusion

Whether ALife is difficult, dangerous, or merely delayed, the Great-Silence framing reframes hypothesised filters as research questions: what computational substrates make abiogenesis cheaper, what formal criteria certify open-endedness, what governance keeps the socio-technological filter benign. Halting ALife research is not the answer; careful, well-instrumented exploration is more likely to de-risk it than neglect.

And — from a broader perspective — the simulation hypothesis hints that ALife may already exist; in some sense, we ourselves could be the “soft” artificial entities the question is about.

References

If you know of additional sources, please open an issue on the repository.

This project

• Karelin, G. (2025). The Fermi Paradox in Artificial Life. In ALIFE 2025: Ciphers of Life — Proceedings of the Artificial Life Conference 2025, Kyoto, Japan, 6–10 October 2025, p. 11. ASME. DOI: 10.1162/ISAL.a.839

Core: Fermi paradox, Great Filter, and astrobiology

• Hart, M. H. (1975). An Explanation for the Absence of Extraterrestrials on Earth. QJRAS.
• Tipler, F. J. (1980). Extraterrestrial Intelligent Beings Do Not Exist. QJRAS.
• Hanson, R. (1998). The Great Filter — Are We Almost Past It?
• Webb, S. (2002). If the Universe Is Teeming with Aliens… Where Is Everybody? Springer.
• Davies, P. (2010). The Eerie Silence: Renewing Our Search for Alien Intelligence.
• Ćirković, M. M. (2018). The Great Silence: Science and Philosophy of Fermi’s Paradox. Oxford UP.
• Snyder-Beattie, A., Sandberg, A., Drexler, K. E., Bonsall, M. (2021). The Timing of Evolutionary Transitions Suggests Intelligent Life Is Rare.
• Mills, D. et al. (2025). A Reassessment of the “Hard Steps” Model.
• Bostrom, N. (2013). Anthropic Bias: Observation Selection Effects.
• Sandberg, A., Armstrong, S., Ćirković, M. (2017). That Is Not Dead Which Can Eternal Lie: The Aestivation Hypothesis.

ALife / SETI bridge

• Smith, H. B., & Sinapayen, L. (2024). Planetary Scale Replication as an Agnostic Biosignature. Proc. ALIFE 2024.
• Wong, M. L., & Bartlett, S. (2022). On the Trajectories of Planetary Civilizations: Asymptotic Burnout vs. Homeostatic Awakening. Proc. ALIFE 2022.
• Furukawa, H., & Walker, S. I. (2018). Major Transitions in Planetary Evolution. Proc. ALIFE 2018.
• Lupisella, M. L. (2004). Using Artificial Life to assess the typicality of terrestrial life. Advances in Space Research, 33(8).
• Bailey, J. et al. (2023). AI as a Great Filter for Astrobiology.
• Garrett, M. A. (2024). Is Artificial Intelligence the Great Filter that Makes Advanced Technical Civilisations Rare in the Universe? Acta Astronautica.

ALife: replicators, open-endedness, and digital abiogenesis

• Sayama, H. (2024). Self-Replication and Open-Ended Evolution in Artificial Life.
• Adami, C. (2024). Evolution and Digital Abiogenesis.
• Ackley, D. H. (2016). Indefinitely Scalable Computing & Artificial Life.
• Packard, N. et al. (2019). An Overview of Open-Ended Evolution.
• Lloyd, S. (2000). Ultimate Physical Limits to Computation. Nature.
• Baltieri, M. et al. (2023). Soft, hard, wet — and now hybrid: the changing computational substrates of ALife.
• Faldor, A. et al. (2024). Toward Quality-Diversity Discovery of Artificial Self-Replicators.

Essays and commentary

• Rees, Sir Martin (UK Astronomer Royal). Why First Contact Could Be With Artificial Life. SETI League editorial — setileague.org/editor/Rees.htm
• Rees, M., & Livio, M. (2024). Most Aliens May Be Artificial Intelligence, Not Life as We Know It. Scientific American — argues that organic human-level intelligence may be a brief phase before the machines take over; post-human, electronic intelligence need not be Darwinian or expansionist and could live quiet, contemplative lives for billions of years (a key target of the dawnstrata reply below). scientificamerican.com/article/most-aliens-may-be-artificial-intelligence-not-life-as-we-know-it
• dawnstrata (15 August 2025). Rare AI and the Fermi Paradox. LessWrong — argues that either technological-singularity events are lethally self-terminating to both their creators and their AI creations, or that humanity is the first intelligent species in galactic history to reach this threshold (a “Rare Intelligence” solution to the Fermi paradox); engages directly with Rees & Livio (2024) on whether AI could be aggressive enough to be fatal to its creators yet also completely non-expansionist after that point. lesswrong.com/posts/jJPTRextDe3chGjho/rare-ai-and-the-fermi-paradox

Workshops and symposia

• Synthesizing Existence: ALife, AI, and the Fermi Paradox Workshop (2023).
• Exploring Exoplanets: The Search for Extraterrestrial Life and Post-Biological Intelligence — international symposium (2015).

Authors with broadly relevant work

• David Kipping — exoplanets and Bayesian reasoning about the emergence of life.
• Clément Vidal — cosmological evolution, high-energy astrobiology.
• Milan M. Ćirković — the Fermi paradox, post-biological evolution, observation-selection effects.

Contribute

If you know of a paper, talk, workshop, or book chapter that belongs in the reference list, please add it as an issue on GitHub with the citation and (ideally) a link to the source.

You can also share this page with the QR code below:

---

This site is a living document and is under active construction. Last updated: 18 May 2026, 05:08 UTC.

← Back to karegeo.github.io