The rapid integration of artificial intelligence (AI) into society has surfaced systemic risks, particularly deep polarization exacerbated by algorithms optimizing for individual engagement. The dominant alignment paradigm, reliant on unilateral control (e.g., RLHF), is ill-suited to address these emergent collective dynamics. This paper proposes a philosophical and computational shift toward symbiotic alignment (SA), moving beyond top-down constraints to a framework of mutual adaptation and co-evolution. We ground SA in collective predictive coding (CPC), reframing human-AI symbiosis as participation in a symbol emergence system (SES). Mathematically, we formalize this interaction as multi-agent reinforcement learning (MARL) augmented by a collective regularization term, driving agents to minimize collective free energy (CFE) while preserving individual autonomy. Crucially, this formulation reveals that social coherence does not require uniformity; within this framework, we computationally reinterpret "plurality" as a stable multimodal distribution of shared beliefs, where diverse worldviews coexist through mutual negotiation. We conclude by outlining the research agenda for realizing this vision: designing AI agents capable of co-creative learning and social mechanisms ("gardeners") that foster trust, thereby steering our technological future toward a flourishing plurality.
Living systems must preserve organization while their components relax, diffuse, or fluctuate. Most origin-of-life accounts place this problem in space, emphasizing membranes and compartments, but persistence also has a temporal dimension. Here we ask whether a dissipative system with delayed feedback can retain statistically meaningful dependence on its own past without an explicit clock or dedicated memory variable. We studied a minimal stochastic delay equation with inhibitory delayed feedback and quantified temporal self-retention by comparing autocorrelation decay time with the intrinsic dissipative timescale. The index crossed unity at τ ≈ 1.918 and remained above unity up to τ = 30.0 under the default parameters. The transition occurred across feedback strengths β = 1.0-3.5, collapsed under nondimensional γτ scaling when β/γ and σ/√γ were fixed, and was distinct from the analytical Hopf boundary. Delay dispersion did not abolish the near-threshold regime, whereas high noise shifted it slightly below unity. Mutual information confirmed persistent dependence between present and delayed states in the long-delay regime. These results identify a minimal delay-dissipative regime in which temporal organization persists beyond dissipation alone, providing a proof of concept for temporal self-retention in artificial-life systems.
Brain-body co-optimization remains a challenging problem, despite increasing interest from the community in recent years. To understand and overcome the challenges, we propose exhaustively mapping a morphology-fitness landscape to study it. To this end, we train controllers for each feasible morphology in a design space of 1,305,840 distinct morphologies, constrained by a computational budget. First, we show that this design space constitutes a good model for studying the brain-body co-optimization problem, and our attempt to exhaustively map it roughly captures the landscape. We then proceed to analyze how evolutionary brain-body co-optimization algorithms work in this design space. The complete knowledge of the morphology-fitness landscape facilitates a better understanding of the results of evolutionary brain-body co-optimization algorithms and how they unfold over evolutionary time in the morphology space. This investigation shows that the experimented algorithms cannot consistently find near-optimal solutions. The search, at times, gets stuck on morphologies that are sometimes one mutation away from better morphologies, and the algorithms cannot efficiently track the fitness gradient in the morphology-fitness landscape. We provide evidence that experimented algorithms regularly undervalue the fitness of individuals with newly mutated bodies and, as a result, eliminate promising morphologies throughout evolution. Our work provides the most concrete demonstration of the challenges of evolutionary brain-body co-optimization. Our findings ground the trends in the literature and provide valuable insights for future work.
Music is an inherently social activity that allows people to share experiences and feel connected with one another. There has been little progress in designing artificial partners exhibiting a similar social experience as playing with another person. Neural network architectures that implement generative models, such as large language models, are suited for producing musical scores. Playing music socially, however, involves more than playing a score; it must complement the other musicians' ideas and keep time correctly. We addressed the question of whether a convincing social experience is made possible by a generative model trained to produce musical scores, not necessarily optimized for synchronization and continuation. The network, a variational autoencoder trained on a large corpus of digital scores, was adapted for a timed call-and-response task with a human partner. Participants played piano with a human or artificial partner-in various configurations-and rated the performance quality and first-person experience of self-other integration. Overall, the artificial partners held promise but were rated lower than human partners. The artificial partner with simplest design and highest similarity parameter was not rated differently from the human partners on some measures, suggesting that interactive rather than generative sophistication is important in enabling social AI.
Neural cellular automata (NCA) are a class of cellular automata where the update rule is parameterized by a neural network that can be trained using gradient descent. In this article, we focus on NCA models used for texture synthesis, where the update rule is inspired by partial differential equations (PDEs) describing reaction-diffusion systems. To train the NCA model, the spatiotemporal domain is discretized, and Euler integration is used to numerically simulate the dynamics. Crucially, it is unclear whether a ground-truth PDE even exists for the task, and NCA training only supervises the final steady state without any trajectory supervision, leaving it an open question whether a trained NCA truly learns continuous dynamics or merely overfits the discretization used during training. We study NCA models at the limit where space-time discretization approaches continuity. We find that existing NCA models tend to overfit the training discretization, especially in the proximity of the initial condition, also called a "seed." To address this, we propose a solution that utilizes uniform noise as the initial condition. We demonstrate the effectiveness of our approach in preserving the consistency of NCA dynamics across a wide range of spatiotemporal granularities. We further show that the resulting model is robust to a stochastic updating scheme and modest additive Gaussian noise. Our improved NCA model enables two new test-time interactions by allowing continuous control over the speed of pattern formation and the scale of the synthesized patterns. We demonstrate this new NCA feature in our interactive online demo. Our work reveals that NCA models can learn continuous dynamics and opens new avenues to studying NCA as a class of PDEs and from a dynamical system's perspective.
The two-dimensional Turing machine is a promising but under used simulation tool for Artificial Life. Single-state 2-D Turing machines exhibit a variety of interesting behaviors, some of which have already been explored. Multistate 2-D Turing machines, despite their potential for simulating even more diverse behaviors, have received little attention to date. We demonstrate the potential of such automata for studying biological phenomena by showing how they can be used to simulate self-similar growth, the spread of disease, and self-reproduction. Some of the results presented here are from investigations that were performed around the time of Dewdney (1989), but they have not been published until now.
An analysis of the language we use in scientific practice is critical to developing more rigorous and sound methodologies. This article argues that how certain methods of description are commonly employed in cognitive science risks obscuring important features of an agent's cognition. We propose to make explicit a method of description whereby the concept of cognitive distinctions is the core principle. A model of referential communication is developed and analyzed as a platform to compare methods of description. We demonstrate that cognitive distinctions, realized in a graph theoretic formalism, better describe the behavior and perspective of a simple model agent than other, less systematic or natural language-dependent methods. We then consider how different descriptions relate to one another in the broader methodological framework of minimally cognitive behavior. Finally, we explore the consequences of, and challenges for, cognitive distinctions as a useful concept and method in the tool kit of cognitive scientists.
Central to the Artificial Life endeavor is the creation of artificial systems that spontaneously generate properties found in the living world, such as autopoiesis, self-replication, evolution, and open-endedness. Though numerous models and paradigms have been proposed, cellular automata (CA) have taken a very important place in the field, notably because they enable the study of phenomena like self-reproduction and autopoiesis. Continuous CA like Lenia have been shown to produce lifelike patterns reminiscent, from both aesthetic and ontological points of view, of biological organisms we call "creatures." We propose Flow-Lenia, a mass conservative extension of Lenia. We present experiments demonstrating its effectiveness in generating spatially localized patterns with complex behaviors and show that the update rule parameters can be optimized to generate complex creatures showing behaviors of interest. Furthermore, we show that Flow-Lenia allows us to embed the parameters of the model, defining the properties of the emerging patterns, within its own dynamics, thus allowing for multispecies simulation. Using the evolutionary activity framework and other metrics, we shed light on the emergent evolutionary dynamics taking place in this system.
PCs are complicated. Yet, being generally more effective, they have replaced typewriters in everyday life. Because of their complications, many of us wonder at PCs as if they were mysterious ghosts in the machine: entities with powers we cannot explain or control, almost supernatural. I analyze how this increase in technological complication may be limiting our society at two levels, one economic and one scientific, and I discuss how the field of Artificial Life (ALife) can attempt to rescue it. At the economic level, there is evidence that computers, being complicated, slow labor productivity rather than increasing it (e.g., maintenance, malware, distractions). Computers are also the subject of debate surrounding technological unemployment and elite overproduction. I advocate for ALife to focus on minimally intrusive developments to our everyday work and to occupy unfilled economic niches, like xenobots or bacterial biofilms. At the scientific level, the surge in artificial intelligence has resulted in many complex algorithms that mimic the cognition happening in brains: Even their creators struggle to make sense of them. I advocate for ALife to focus more on basal forms of cognition, cognition that requires as little "brain" as possible, potentially none-algorithms that think through their bodies, stripped of any superfluous complications, just like typewriters. Ultimately, my goal is for the reader to ask themselves what values should drive ALife.
Camouflage in nature seems to arise from competition between predator and prey. To survive, predators must find prey, while prey must avoid being found. A simulation model of that adversarial relationship is presented here. Camouflage patterns of prey coevolve in competition with visual perception of predators. During their lifetimes, predators learn to better locate the camouflaged prey they encounter. The environment for this 2-D simulation is provided by photographs of natural scenes. The model consists of two evolving populations, one of prey and another of predators. Conflict between these populations produces both effective prey camouflage and predators able to "break" camouflage. The resulting open-source Artificial Life model can help the study of camouflage in nature and the perceptual phenomenon of camouflage more generally.
This article presents Benefit Game 2.0, a multiscreen Artificial Life gameplay installation. Saccharina latissima, a seaweed species economically beneficial to humans but threatened by overexploitation, motivates the creation of this artwork. Technically, the authors create an underwater virtual ecosystem consisting of a seaweed swarm and symbiotic fungi, created using procedural content generation via machine learning and rule-based methods. Moreover, the work features a unique cybernetic loop structure, incorporating audience observation and game token interactions. This virtual system is also symbolically influenced in real time by indoor carbon dioxide measurements, serving as an artistic metaphor for the broader impacts of climate change. This integration with the physical game machine underscores the fragile relationship between human activities and the environment under severe global climate change and immerses the audience in the challenging balance between sustainability and profit seeking in this context.
This article introduces the System 0/1/2/3 framework as an extension of dual-process theory, employing a quad-process model of cognition. Expanding upon System 1 (fast, intuitive thinking) and System 2 (slow, deliberative thinking), we incorporate System 0, which represents precognitive embodied processes, and System 3, which encompasses collective intelligence and symbol emergence. We contextualize this model within Bergson's philosophy by adopting multiscale time theory to unify the diverse temporal dynamics of cognition. System 0 emphasizes morphological computation and passive dynamics, illustrating how physical embodiment enables adaptive behavior without explicit neural processing. Systems 1 and 2 are explained from a constructive perspective, incorporating neurodynamical and artificial intelligence (AI) viewpoints. In System 3, we introduce collective predictive coding to explain how societal-level adaptation and symbol emergence operate over extended timescales. This comprehensive framework ranges from rapid embodied reactions to slow-evolving collective intelligence, offering a unified perspective on cognition across multiple timescales, levels of abstraction, and forms of human intelligence. The System 0/1/2/3 model provides a novel theoretical foundation for understanding the interplay between adaptive and cognitive processes, thereby opening new avenues for research in cognitive science, AI, robotics, and collective intelligence.
The NK fitness landscape is a well-known model with which to study evolutionary dynamics in landscapes of different ruggedness. However, the model is static, and genomes are typically small, allowing observations over only a short adaptive period. Here we introduce an extension to the model that allows the experimenter to set the velocity at which the landscape changes independently from other parameters, such as the ruggedness or the mutation rate. We find that, similar to the previously observed complexity catastrophe, where evolution comes to a halt when environments become too complex due to overly high degrees of epistasis, here the same phenomenon occurs when changes happen too rapidly. Our expanded model also preserves essential properties of the static NK landscape, allowing for proper comparisons between static and dynamic landscapes.
The Self-Optimization (SO) model can be considered as the third operational mode of the classical Hopfield Network, leveraging the power of associative memory to enhance optimization performance. Moreover, it has been argued to express characteristics of minimal agency, which renders it useful for the study of artificial life. In this article, we draw attention to another facet of the SO model: its capacity for creativity. Drawing on creativity studies, we argue that the model satisfies the necessary and sufficient conditions of a creative process. Moreover, we show that learning is needed to find creative outcomes above chance probability. Furthermore, we demonstrate that modifying the learning parameters in the SO model gives rise to four different regimes that can account for both creative products and inconclusive outcomes, thus providing a framework for studying and understanding the emergence of creative behaviors in artificial systems that learn.
With the recent Nobel Prize awarded for radical advances in protein discovery, foundation models (FMs) for exploring large combinatorial spaces promise to revolutionize many scientific fields. Artificial Life (ALife) has not yet integrated FMs, thus presenting a major opportunity for the field to alleviate the historical burden of relying chiefly on manual design and trial and error to discover the configurations of lifelike simulations. This article presents, for the first time, a successful realization of this opportunity using vision-language FMs. The proposed approach, called automated search for Artificial Life (ASAL), (a) finds simulations that produce target phenomena, (b) discovers simulations that generate temporally open-ended novelty, and (c) illuminates an entire space of interestingly diverse simulations. Because of the generality of FMs, ASAL works effectively across a diverse range of ALife substrates, including Boids, Particle Life, the Game of Life, Lenia, and neural cellular automata. A major result highlighting the potential of this technique is the discovery of previously unseen Lenia and Boids life-forms, as well as cellular automata that are open-ended like Conway's Game of Life. Additionally, the use of FMs allows for the quantification of previously qualitative phenomena in a human-aligned way. This new paradigm promises to accelerate ALife research beyond what is possible through human ingenuity alone.
Robot controllers are often optimized for a single robot in a single environment. This approach proves brittle, as such a controller will often fail to produce sensible behavior for a new morphology or environment. In comparison, animal gaits are robust and versatile. By observing animals, and attempting to extract general principles of locomotion from their movement, we aim to design a single, decentralized controller applicable to diverse morphologies and environments. The controller implements the three components of (a) undulation, (b) peristalsis, and (c) leg motion, which we believe are the essential elements in most animal gaits. This work is a first step toward a general controller. Accordingly, the controller has been evaluated on a limited range of simulated centipede-like robot morphologies. The centipede is chosen as inspiration because it moves using both body contractions and legged locomotion. For a controller to work in qualitatively different settings, it must also be able to exhibit qualitatively different behaviors. We find that six different modes of locomotion emerge from our controller in response to environmental and morphological changes. We also find that different parts of the centipede model can exhibit different modes of locomotion, simultaneously, based on local morphological features. This controller can potentially aid in the design or evolution of robots, by quickly testing the potential of a morphology, or be used to get insights about underlying locomotion principles in the centipede.
