Organisms. Journal of Biological Sciences

Special Issue, “What AI Can Learn from Biology”

2025-12-20T22:31:12+00:00

This special issue uses AI to cast light on the nature of life. Many assume that life emerges from a blend of information and complexity. If this is the case, then we might expect a future generation of machines to exhibit lifelike behavior or, as some would claim, to come alive. Two perspectives are offered for considering the question of life: agency and intelligence. Intelligence is associated with information, rationality and consequent knowledge representations, while agency associates with embodiment, judgment and material organization. Predictions about machine life rely on conceptions of intelligence, but the addition of agency to the analysis of life and lifelike behavior results in nuanced conclusions that can beneficially inform regulation and future research.

Keywords: AI, cognition, worlding, thermodynamics

AI in This World and the Next

2025-10-17T19:38:46+00:00

As the symptoms of our self-inflicted planetary emergency become ever more alarming, hope seems to be growing that AI technologies can make our capitalist way of life more sustainable. Some even believe that machine intelligence will avert impending catastrophe more or less by itself. But the evidence of history should caution us against such heady Promethean optimism. Millennia of human experience suggest that only radical systemic change can halt our perilous trajectory. AI interventions and other such modern techno-fixes will simply not be enough.

An exciting new theoretical paradigm in the humanities and social sciences can help us grasp the full urgency of this message from history. Briefly stated, it recasts reality itself as a variable relational effect, one that humans co-produce with non-humans in the course of their everyday life practices. And just as practices have varied widely over time and space, so life has come to be experienced in a “pluriverse” of many different worlds, not in a universe of just one. An alternative pluriversal vision of history then allows us to identify striking correspondences between the sustainability of communities and their particular ways of “worlding.”

Most immediately, one can correlate the consistent sustainability of non-modern communities, past and present, with their commitment to living by a common set of metaphysical principles or “laws of being.” In stark contrast, the technoscientifc capitalist world of our own modernity, a world that current AI practices are hard-wired to perpetuate, directly violates all of these same tried-and-tested laws. The dire ecological consequences for the planet are now all too plain to see. It is vital that we learn lessons from the vast inventory of non-modern experiences and commit to re-engineering our way of worlding along more ecologically reponsible lines. Modified forms of AI can absolutely help us to realize a more livable future world in practice. But they cannot save us all by themselves.

Modelling the Threat from AI: Putting Agency on the Agenda

2025-12-20T22:42:49+00:00

The AI existential-risk narrative focuses on an ‘intelligence explosion’ leading to uncontrollable superintelligence. This paper contends that the more plausible and proximate threat is the emergence of strong biological-style agency in digital systems, independent of high intelligence. Drawing on systems biology and thermodynamics, it contrasts mechanistic with organic agency: living organisms are autocatalytic systems that harness environmental energy for self-maintenance and reproduction, whereas current Autonomous/Intelligent Systems pursue only externally assigned goals. Evolution produced robust agency in bacteria, slime molds, and insects long before cognition. Recent work in embodied neural networks and bio-inspired computing shows that complex adaptive behavior can arise in machines through structural coupling with their environment that occurs without symbolic reasoning. Deliberate or accidental development of energy-seeking, self-reproducing ‘biodigital agents’ could therefore yield invasive, unpredictable systems well below superintelligent levels. The paper advocates shifting AI safety priorities from anthropomorphic ethics and alignment to measurable biophysical criteria derived from the definition of life. Recommended measures include engineering standards prohibiting direct environmental energy harvesting by A/IS, global energy audits to detect emergent agency, and epidemiological containment frameworks—thereby preventing a Cambrian-like explosion of machine agency before superintelligence becomes feasible.

Keywords: AI, superintelligence, intelligence explosion, biodigital agents

Could Artificial Intelligence (AI) Become a Responsible Agent: Artificial Agency (AA)?

2025-12-20T22:55:06+00:00

Responding to concerns that superintelligent AI could escape human control, this paper argues that the true existential question is not intelligence but agency, and that artificial intelligence as currently conceived poses no threat of responsible agency. Intelligence can be fully artificial and beneficial (books, databases, algorithms) without ever bearing responsibility. Responsibility belongs exclusively to agents, specifically biological agents. Biological agency requires causal independence, intentionality, creativity, and above all the active harnessing of stochasticity to generate novel, goal-directed behavior that is neither predetermined nor merely random. Organisms achieve this at every level—from ion channels and immune-system hypermutation to neural decision-making and social anticipation—by constraining chance rather than eliminating it. Choice in living systems resembles poker rather than chess: iterative, intuitive, socially embedded, and inherently unpredictable even in principle. Algorithmic systems, even those incorporating randomness, cannot replicate this multi-level process. Creating genuine artificial agency would demand reproducing biology’s constrained use of stochasticity across scales. Only then could a machine become a responsible (or irresponsible) agent. If achieved, the distinction between living and artificial would collapse, raising profound ethical questions. Until then, the risk lies not in AI itself but in failing to regulate research that might inadvertently cross this threshold.

Keywords: biological agents, stochasticity, responsibility, intentionality

Could Machines Develop Autonomous Agency?

2025-12-20T23:10:59+00:00

“Could machines develop autonomous agency?” To address this question, we explored the recent return of the concept of agency in biological discourse. At the end of the 19th century, the successful development of physics and chemistry motivated some biologists to adopt a physicalist stance, positing that biology can be reduced to physics and chemistry. This theoretical approach became dominant during the 20th century with the advent of molecular biology while teleology, agency and normativity disappeared from the biological lexicon. The failure of molecular biology to explain complex biological organization probably led to the reintroduction of these concepts in the biological sciences and philosophy of biology. In addition to the historicity of organisms (they are the product of organismal reproduction throughout phylogenesis), the intrinsic properties of biological objects are linked to the precariousness of life as exemplified by the need to search for food and to avoid being eaten. Moreover, the continuous need to counteract entropy also involves the capacity of organisms to synthesize their own chemical components and reproduce. From this historical narrative, we conclude that it is unlikely that machines could develop minimal intrinsic agency. On the contrary, when they appear to express agency, it is of external origin, reflecting the agency of the humans that created such machines.

Keywords: teleology, historicity, goal-directedness, natural agency, artificial agency

Chess as a Model of Collective Intelligence

2025-06-13T12:49:30+00:00

Chess is a much-studied virtual world in which human and artificially-intelligent players move pieces toward desired ends, within established rules. The typical scenario involves top-down control where a single cognitive agent plans and executes moves using the pieces as its embodiment within the chess universe. However, ultimately both biological and engineered agents are composed of parts, with radically differing degrees of competency. The emerging field of Diverse Intelligence seeks to understand how coherent behavior and goal-directed navigation of problem spaces arises in compound agents from the interaction of their simpler components. Thus, we explored the world of chess rules from the perspective of collective intelligence, and characterized a bottom-up version of this classic game in which there is no central controller or long-term planning. Rather, each individual piece has its own drives and makes decisions based on local, limited information and its own goals. We analyzed the behavior of this distributed agent when playing against Stockfish, a standard chess algorithm. We tested a few individual policies designed by hand, and then implemented an evolutionary algorithm to see how the individuals’ behavioral genomes would evolve under selection applied to the chess-based fitness of the collective agent. We observed that despite the minimal intelligence of each piece, the team of distributed chess pieces exhibit Elo of up to ~1050, equivalent to a novice human chess player. And, compared to advanced chess engines like Stockfish, the distributed chess pieces are significantly more efficient in computing. Distributed chess pieces select their next move approximately 7 times faster than the Stockfish Engine with a search depth of 8. Investigating different local policies for the distributed agents, we found that policies promoting offense, such as swarming the opposing king and opposing highest valued piece, moving less cautiously, and a radius of vision of 4 spaces yields optimal performance. Comparisons between centralized and distributed versions of familiar minimal environments have the potential to shed light on the scaling of cognition and the requirements for collective intelligence in naturally evolved and engineered systems.

Keywords: decentralized intelligence, emergence, behavior, minimal models, distributed systems

What Drives the Brain? Organizational Changes, FEP and Anti-entropy

2025-12-20T23:24:09+00:00

he free-energy principle (FEP) provides a computational, physical and teleological theory for understanding biological organization as cognitive agent minimizing their entropy in relation to their environment. Is minimizing entropy the first principle driving all dynamics of cognition? Is it enough to account for organizational changes in an open-ended way? After a general presentation of the literature on the FEP, we turn to the paradoxical case of the brain under the influence of psychedelics, where the FEP is challenged by an increased cerebral entropy, which induces organizational changes of the cognition. Building on this paradox, we identify some limits of the FEP, notably applying concepts of information, optimization and predefined phase space to biology that do not fit our criteria for a theory of biological organization. We also identify two aspects of entropy in physics and in the FEP: the local entropic trend that implies variations and the global entropic trend that leads to homogeneization and stability. Extending these concepts outside of their physicomathematical context, we contribute to an organicist theoretical alternative where living systems find a balance between these two trends, and, conceptually, a biological system’s disorganization enable its “unprestatable” reorganization and so its open-ended evolution.

Keywords: free energy principle, anti-entropy, entropy, novelty, biological organization, organicism

The Rate of Entropy Production as a Lyapunov Function in Biophysical-chemical Systems

2025-10-19T08:07:03+00:00

An overview of the link between nonequilibrium thermodynamics and complexity theory is offered here, showing how the entropy production rate can be quantified through the spectrum of the Lyapunov exponents. The work shows how the entropy production per unit of time meets the necessary and sufficient conditions to be a Lyapunov function and constitutes per se an extremal principle. The entropy production fractal dimension conjecture is also established. The work demonstrates how the rate of entropy production as a non-extremal criterion represents an alternative way for sensitivity analysis of differential equations. Finally, in an extension to biophysical-chemical systems, on the one hand, the study presents the use of the dissipation function as a thermodynamic potential out of equilibrium in the characterization of biological phase transitions. On the other hand, it evidences that the entropy production rate represents a physical quantity that can be used to evaluate the complexity and robustness of cancer.

Keywords: nonequilibrium thermodynamics, entropy production rate, Lyapunov function, complexity, biological phase transitions