The story begins with Cordeschi’s original insight, which Gagliardi revisits and reinterprets. In the 1930s and 1940s, researchers in the emerging behavioral sciences began constructing machines—mechanical and, later, electronic artifacts—that could reproduce intelligent and adaptive behavior. These were not merely engineering curiosities. They functioned as scientific models: ways of testing hypotheses about how minds and nervous systems might work. Kenneth Craik’s 1943 proposal that organisms carry internal models of external reality, Warren McCulloch and Walter Pitts’s 1943 logical calculus of neural activity, and the later cybernetic machines of Norbert Wiener’s circle all exemplified what came to be known as the “Synthetic Method.” Rather than analyzing a natural system into its components and deriving its behavior from first principles, the synthetic method proceeds in the opposite direction: it assembles an artifact from simple parts and observes whether the target behavior emerges.

Gagliardi’s central move is to pair this familiar narrative with a less obvious one from physics. In 1920, Wilhelm Lenz proposed a radically simplified model of ferromagnetism, and his student Ernst Ising worked out its properties in 1925. The Lenz–Ising model represents a magnetic material as a lattice of spins, each of which can point up or down and interacts only with its nearest neighbors. Nothing about the model resembles a real magnet in its fine detail. Yet this deliberately crude artifact turned out to capture something essential about how collective phenomena—phase transitions, critical points, cooperative behavior—arise from local interactions. Gagliardi argues that the introduction of the Ising model in the 1920s represents a methodological turning point in the physical sciences directly analogous to the one occurring at the same time in the behavioral sciences: the embrace of artificial, simulative models as legitimate instruments of scientific understanding.

Why did both fields converge on this strategy? Gagliardi’s answer lies in the mathematics of complexity. In the kinetic theory of gases, the nineteenth-century triumph of Ludwig Boltzmann and James Clerk Maxwell, it was possible to move analytically from the microscopic behavior of particles to macroscopic laws: the model could be solved in closed form, yielding the ideal gas law. But the Ising model resists such treatment. It has been formally proven, in work by Sorin Istrail published in 2000, that computing the ground state of the Ising model is computationally intractable—an NP-hard problem. Assuming the widely held conjecture that P does not equal NP, no closed-form analytical solution exists except in trivial cases. The same wall of intractability confronts cognitive modelers: Paul Thagard and Kevin Verbeurgt showed in 1998 that a connectionist model of coherence as constraint satisfaction is equivalent to the NP-complete Max-Cut problem. When exact analysis is impossible, the only route to understanding is simulation—running the model and observing its behavior.

This is where the concept of computational irreducibility enters. For complex systems, whether biological or material, there is often no shortcut from the model’s specification to its outcomes; one must simply let the dynamics unfold. Gagliardi contends that this epistemic necessity, rather than any mere fashion or convenience, explains why both cognitive science and statistical physics became what he calls “complexity sciences.” Both disciplines, in his phrase, “discovered the artificial”: they came to accept that understanding a system may require building an internal model of it whose behavior can be observed, even when no analytic derivation of that behavior is available. The artificial model becomes not a substitute for explanation but the very medium of explanation.

The philosophical stakes of this claim are considerable. Traditional accounts of scientific explanation, descending from Galileo’s famous declaration that the book of nature is written in mathematical language, privilege derivation: to explain is to deduce consequences from mathematical first principles. Gagliardi argues that the synthetic method extends rather than abandons this Galilean language. The new language of models and simulations adds to mathematical description the capacity to create and run artificial systems whose behavior can be studied. He points to Ernst Mach’s nineteenth-century observation that all science seeks to replace or economize experience through the mental reproduction of facts—reproductions that are easier to handle than experience itself and can stand in for it. Simulation, on this view, is the modern technological fulfillment of Mach’s epistemology of thought-economies, closer in spirit to a gedankenexperiment than to a laboratory measurement.

The article also engages a live debate in the philosophy of science about the epistemic status of computer simulations. Are simulations experiments, or are they theory? Gagliardi notes that observing the physical system under study belongs to the empirical verification phase of the scientific method, whereas observing a simulation pertains to theory and to the deduction of a model’s properties. He acknowledges that this distinction is contested: philosophers such as Anouk Barberousse, Cyrille Imbert and Sara Franceschelli, Claus Beisbart, and Judith Jebeile have argued for a genuine affinity between experiments and simulations, while others, including Julian McClelland and Darrell Rowbottom, treat simulations as instruments rather than faithful recreations of phenomena. Gagliardi’s historical framing gives this debate new context: the question of what simulations are is inseparable from the century-long process by which both physics and the mind sciences learned to trust artificial models.

Perhaps the most provocative element of the paper is its appeal to weak emergence, a concept developed by Mark Bedau, to characterize the new kind of scientific understanding that the synthetic method affords. In systems like the Ising model, there is no causal reductionism linking the internal model to macroscopic phenomena in the way that the kinetic theory analytically links molecular motion to the gas laws. Instead, macroscopic patterns depend on, but are not derivable from, the micro-dynamics—a dependence that must be exhibited through simulation rather than demonstrated through derivation. Gagliardi suggests that explanation centered on the internal functional organization of systems, in the tradition of Robert Cummins’s functional analysis, offers a shared explanatory idiom for both disciplines. Understanding becomes a matter of seeing how a system’s organization produces its capacities, a form of comprehension that simulation makes possible and that pure mathematics alone cannot deliver.

The convergence Gagliardi describes has contemporary resonance. The 2024 Nobel Prize in Physics, awarded for foundational work on artificial neural networks, underscored how deeply the physics of collective phenomena and the science of mind have become intertwined: John Hopfield’s 1982 neural networks drew explicitly on statistical-mechanical ideas, and Boltzmann machines, developed by David Ackley, Geoffrey Hinton and Terrence Sejnowski in 1985, carry the name of the great statistical physicist. Recent work by Iris van Rooij and colleagues has argued for reclaiming artificial intelligence, with its computational models and simulation techniques, as a theoretical foundation for cognitive science, with computational complexity theory playing a central role. Gagliardi’s historical analysis provides the deep background for these developments: they are not accidents of interdisciplinary fashion but expressions of a shared “Culture of the Artificial” that has been consolidating for a hundred years.

The paper, dedicated to the memory of Roberto Cordeschi, ultimately offers a vision of scientific unity that is neither reductionist nor pluralist in the usual senses. Cognitive science and statistical physics remain distinct disciplines with distinct subject matters, yet they are aligned at a deeper epistemic level: both have learned that the road to understanding complex, adaptive, collective phenomena runs through the construction of artificial models and the observation of their simulated behavior. In extending the Galilean language of mathematics into a language of model-building and simulation, twentieth-century science did not abandon rigor; it redefined what counts as explanation for a world whose systems are too complex to be solved, and rich enough to be understood only from the inside out.

Subject of Research: The historical convergence of the synthetic method and artificial modeling in cognitive science and statistical physics

Article Title: The discovery of the artificial and the use of the synthetic method in cognitive and physical sciences

Article References: Gagliardi, F. (2026). The discovery of the artificial and the use of the synthetic method in cognitive and physical sciences. AI & SOCIETY. https://doi.org/10.1007/s00146-026-03291-4

Image Credits: AI Generated

DOI: 10.1007/s00146-026-03291-4

Keywords: synthetic method, discovery of the artificial, cognitive science, statistical physics, Ising model, cybernetics, computer simulation, computational irreducibility, weak emergence, complexity science, Roberto Cordeschi, history of science

The authors open with a historical contrast that has already struck a chord with readers. As Walter Isaacson describes in his 2007 biography of Albert Einstein, the physicist worked largely alone, with a blackboard, long stretches of uninterrupted thought, and very few external aids. The papers that continue to shape physics emerged from those constrained, unhurried conditions. Einstein’s creative process was slow and often lonely, requiring months or years of sitting with a problem, discarding half-formed ideas, and returning to the same question again and again. Today, by contrast, an academic can summon hundreds of computational helpers at once. These systems calculate, retrieve, rewrite, suggest next steps, and even propose titles and structure arguments. The cost of producing text, tables, and literature reviews has collapsed. Yet the volume of papers, the speed of review cycles, the density of citation networks, and the length of curricula vitae have all grown without a matching surge in works that reorient entire disciplines.

At the heart of the commentary is a distinction the authors keep returning to: AI multiplies the number of assistants, but it does not multiply the capacity for sustained, risky, and inefficient thinking. The technology lowers the cost of being busy. It has not lowered the cost of becoming a genius, or even of protecting the conditions under which difficult thinking sometimes occurs. A researcher can now generate drafts, run analyses, fill notebooks, and produce polished texts at industrial speed. But the willingness to stay with one difficult, possibly fruitless question for a decade remains an expensive proposition. It demands time that looks unproductive, the tolerance to appear slow while everyone else publishes, and institutional space that permits that appearance of idleness. In the authors’ framing, the new tools have made visible productivity easier than ever, and that is precisely the problem.

The commentary reaches back to an earlier technological anxiety to sharpen its point. Norbert Wiener, in his 1961 book Cybernetics, warned that the real danger of automation was not that machines would become too intelligent, but that humans would become too willing to hand over the slow, uncertain parts of thinking. Herwandito and Pramana argue that the current incentive structure proves Wiener right in a way he could not have foreseen. Universities and funding bodies reward productivity, and AI makes productivity trivially easy to demonstrate. The rational response for most researchers is to use the new assistants to clear many small questions quickly rather than to guard the slow, uncertain work of one large question. A hundred researchers, each equipped with a hundred AI helpers, can solve a thousand modest problems in a year. The same group, if it chose, could spend ten years solving a single unreasonable problem. Almost everything in the current system, the authors contend, pushes them toward the first option.

The tension is visible in the everyday habits of working scholars. The authors describe seeing it in their own working lives and in the practices of younger colleagues. The tools are genuinely helpful for clearing routine tasks, and they have freed hours that could, in principle, be spent on harder thinking. In practice, those hours are usually filled with more production: another draft, another literature scan, another incremental paper. The result is an academic culture that feels increasingly efficient and increasingly thin. The community has become very good at generating outputs, the authors write, but less practiced in protecting the conditions under which something genuinely new might appear. The freed time does not flow toward depth; it flows toward volume.

Importantly, the commentary is not a call for a romantic return to chalk and isolation. The authors are explicit that these assistants are useful and that they remove drudgery that never had any intrinsic scientific value. If researchers insisted on it, the time saved by automation could be redirected toward deeper work. The problem, they argue, is that nobody insists. The saved time is spent producing more of the same type of work, only faster. The system rewards that choice, and the tools make it almost effortless to comply. What looks like a technological revolution, in this reading, is functioning as an accelerant for the existing incentive structure rather than a challenge to it.

The authors do offer a path forward, though it is a demanding one. Great works may still appear, they suggest, from people who treat AI the way earlier generations treated libraries or calculating machines: as support for a long, stubborn conversation with a problem that refuses easy answers. That would require a deliberate decision to keep some of the old inefficiency, protecting stretches of time in which nothing publishable is produced, and judging the value of that time by the quality of the questions rather than the quantity of the output. It would also require institutions willing to recognize and reward that kind of patience, a significant ask in a sector increasingly measured by grant income, publication counts, and throughput metrics.

The piece appears under the banner of Curmudgeon Corner, which the journal’s editor describes as a short opinionated column on trends in technology, arts, science, and society, commenting on issues of concern to the research community and the wider public. The column’s framing question is stark: what is it to be human in the age of the AI machine? The editors note that while the drive toward super-human intelligence promises benefits to society, it also raises deep concerns, including existential risk, making an ongoing conversation between technology and society essential. Herwandito and Pramana’s contribution fits squarely in that tradition, using a personal, letter-to-the-editor format to raise a question that quantitative studies of AI-assisted science have so far struggled to answer directly.

Whether the commentary’s diagnosis proves correct is an empirical question that will take years to settle. Historians of science have long debated whether breakthroughs cluster in conditions of constraint or abundance, and the present moment offers an unprecedented natural experiment: an entire research ecosystem suddenly equipped with generative tools, its output measurable in real time. If the authors are right, the telltale sign will not be in the publication statistics, which will almost certainly keep climbing, but in the absence of the rare papers that redirect whole fields. Until such choices about how to spend freed attention become more common, the authors conclude, no one should be surprised that the blackboards of the past cast longer shadows than the crowded dashboards of the present. Humanity has given itself more hands to work with, they write, but has not yet figured out how to give itself more depth.

Subject of Research: The impact of AI research assistants on the conditions for major scientific breakthroughs in academia

Article Title: Einstein had a blackboard. We have a hundred assistants. Where are the great works?

Article References: Herwandito, S., & Pramana (2026). Einstein had a blackboard. We have a hundred assistants. Where are the great works?. AI & SOCIETY. https://doi.org/10.1007/s00146-026-03358-2

Image Credits: AI Generated

DOI: 10.1007/s00146-026-03358-2

Keywords: artificial intelligence, academic research, scientific breakthroughs, AI & Society, research productivity, Albert Einstein, Norbert Wiener, cybernetics, academia, creativity, research incentives, generative AI

Pask’s cybernetics, developed in the 1970s through works such as his 1975 volume on conversation, cognition and learning, treats learning not as the transfer of information from an expert to a novice but as a continuous loop of conversations between participants who must each build and test their own understanding of what the other knows. In this view, a classroom is a self-organizing system of interacting actors, and knowledge emerges from the negotiations among them. When an AI system enters the room, it does not merely deliver content; it becomes a node in that network, sending and receiving feedback, reshaping the conversations around it even as those conversations reshape how it is used. Tilak uses this lens to visualize classroom interactions across all four studies, mapping how human and machine agents circulate questions, answers, and corrections in cycles that never fully settle.

The first study examined teachers using MagicSchool AI, a generative platform aimed at educators, to design course blueprints. Rather than treating the AI as an automatic lesson generator, the participating teachers engaged in an iterative dialogue with it, feeding it their goals, critiquing its drafts, and revising prompts in response to what the system produced. The cybernetic reading of this process emphasizes that the value came from the feedback loop itself: the AI’s outputs forced teachers to articulate their pedagogical intentions more precisely, and those articulations in turn steered subsequent generations toward usable designs. The study, published in the Journal of Sociocybernetics in 2024, framed this as a participatory qualitative analysis, with teachers acting as co-designers of the curriculum rather than passive consumers of machine suggestions.

The second study moved from planning to production, asking college psychology students to use text-to-image generative AI to create comic strip storyboards for their coursework. Storyboarding has a long pedigree as a classroom technique for breaking down complex narratives into manageable visual sequences, and generative image tools gave students a new collaborator in that task. Students had to translate abstract psychological concepts into concrete scene descriptions, evaluate the sometimes uncanny outputs, and negotiate revisions. The synthesis highlights how this workflow distributed cognition across human and machine: the AI contributed visual fluency while the students contributed conceptual grounding, and the comic strips that emerged would have been impossible for either party alone. The findings, also published in the Journal of Sociocybernetics, suggest that creative assignments can channel generative AI away from plagiarism concerns and toward genuine collaborative problem-solving.

The third and perhaps most provocative study cast AI as a Socratic opponent. In a college psychology course, students engaged in interactive discussions where an AI was positioned not as an answer machine but as an interlocutor that challenged their reasoning, much as the classical Socratic method demands. Tilak and colleagues then applied network analysis, a statistical technique for mapping relationships among actors, to the resulting transcripts, revealing patterns in how discourse flowed among students and the machine. Published in 2025 in the Journal of Sociocybernetics, the comparative analysis showed that when the AI’s role was explicitly framed as an opponent rather than an oracle, students treated its claims more critically and defended or revised their own positions more actively. This connects to a growing literature on conversational agents in group settings, including research on how such agents should communicate within group chats, and to recent comparative work suggesting that AI tutors can, under the right conditions, support critical thinking alongside human instructors.

The fourth study addressed an equity concern that most AI-in-education debates overlook: students with disabilities. Working with high schoolers with ADHD in mathematics classes, Tilak and team member Marzena Bogacki built a cybernetic feedback mechanism using ALEKS, the Assessment and Learning in Knowledge Spaces platform. ALEKS models a learner’s knowledge as a network of concepts and continuously probes which items the learner is ready to acquire next. The research reframed this adaptive engine as a feedback controller in Pask’s sense: assessment data cycles back into instruction in real time, allowing the system, the teacher, and the student to adjust one another’s behavior. For learners whose attention and executive function make conventional pacing difficult, this tight loop of observation and adaptation offered a structure that static worksheets could not, though the study also underscored the importance of keeping a human teacher in the loop to interpret the data and respond to needs the algorithm cannot see.

Taken together, the four scenarios illustrate five of the six configurations of collaborative AI use originally outlined by learning scientist Mike Sharples in his 2023 work on social generative AI for education. Sharples has argued that educational AI should move beyond the tutor-as-responder model toward a vision in which AI systems participate in the social construction of knowledge, conversing with students and with each other. Tilak’s synthesis demonstrates what those abstract configurations look like on the ground: AI as a design partner for teachers, as a co-creative medium for students, as a dialectical adversary in discussion, and as an adaptive monitor in special education. Each configuration preserves the human relationships at the center of the classroom while extending what those relationships can accomplish.

The paper is equally candid about the grey areas that experts in AI education research have carved out, including unresolved questions about design and implementation, classroom relationships, and assessment. Ethical debates over trustworthy and educational AI are accelerating globally, and a 2024 consensus statement in the Proceedings of the National Academy of Sciences warned of threats to scientific integrity in the age of generative AI. Tilak’s response is not to retreat from the technology but to give educators a structured way to think about where responsibility lies in a hybrid system. If learning is a conversation, then accountability is also distributed: teachers must interrogate what the AI produces, students must own their interpretive contributions, and developers must build systems transparent enough to participate honestly in those exchanges. The synthesis argues that this distribution only works when it is designed deliberately, not left to emerge by accident.

The practical payoff of the research program is a set of shepherding suggestions for three core activities: AI-mediated curriculum design, classroom problem-solving and information acquisition, and nimble student evaluation. For design, the recommendation is to treat prompt engineering as an extended conversation about pedagogy rather than a one-shot command. For classroom work, the recommendation is to assign AI roles, opponent, collaborator, or monitor, explicitly, so students understand what kind of intelligence they are engaging and calibrate their trust accordingly. For assessment, the recommendation is to use learning data as part of a feedback loop that informs the next teaching move, rather than as a terminal judgment. The author, whose broader program also includes experimental work on sociohistorically sensitive information search tools and conceptual studies linking cybernetics to psychological theory, notes that the sixth of Sharples’s collaborative configurations remains unexplored, and that a planned series of subject-specific scenarios using AI as a collaborative coach is still ahead.

What makes this research resonate beyond its immediate findings is its refusal of the two tired scripts that dominate public discussion of educational AI: utopian salvation and inevitable ruin. Pask’s cybernetics, precisely because it predates the current hype cycle, offers a vocabulary for something in between, a view of classrooms as living systems in which new actors can be integrated without displacing the old ones. As schools worldwide race to draft AI policies with little empirical grounding, the four studies offer evidence that the deciding variable is not the sophistication of the model but the quality of the conversational architecture built around it. In that sense, the most futuristic guidance for AI in education may come from a theorist who never saw a chatbot: design the conversation first, and the technology will find its place within it.

Subject of Research: AI-mediated collaborative learning guided by cybernetic conversation theory in adult and special education classrooms

Article Title: A Cybernetic Guide to Implementing AI for Collaborative Learning: A Synthesis of Four Studies Conducted with Adult Learners

Article References: Tilak, S. (2025). A Cybernetic Guide to Implementing AI for Collaborative Learning: A Synthesis of Four Studies Conducted with Adult Learners. Frontiers of Digital Education, 2(4), Article 37. https://doi.org/10.1007/s44366-025-0074-7

Image Credits: AI Generated

DOI: 10.1007/s44366-025-0074-7

Keywords: artificial intelligence, cybernetics, Gordon Pask, collaborative learning, generative AI, education, special education, conversation theory, Mike Sharples, Socratic method, adaptive learning, classroom design