Machines compute, humans decide: Meaning and judgment remain human in AI age

A new peer-reviewed review published in AI & Society argues that computation is not separate from human reasoning, culture, and ethics, but embedded within them.

In the study titled “Eco-cognitive computationalism,” computer scientist and philosopher Gordana Dodig-Crnkovic examines and evaluates the theoretical framework developed by Lorenzo Magnani in his book Eco-Cognitive Computationalism. The review offers a systematic account of computation as a human, culturally situated activity rather than a purely formal or machine-driven process.

Computation reframed as a human cognitive practice

Eco-cognitive computationalism presents it as a historical and cultural practice through which humans organize uncertainty and generate meaning. The author explains that Magnani treats computation as something humans actively cultivate through interaction with tools, symbols, environments, and technologies.

The review traces how the book grounds this argument in the historical co-development of information, cognition, and environment. Computation emerges not as a detached technical layer but as part of a long process through which humans domesticate raw materials and symbolic systems into supports for reasoning. Information is described as structured difference within a context, while cognition is the human capacity to interpret and stabilize those differences into usable forms. Computation, in this sense, becomes a method for transforming ignorance into intelligibility.

This perspective challenges dominant narratives in artificial intelligence that frame cognitive capacity as something that can be fully extracted, replicated, or optimized in machines. Instead, Magnani’s framework emphasizes that computation always operates within ecological settings shaped by history, culture, and social practice. The author highlights how this view aligns computation with world-making rather than mere problem-solving, stressing that technologies do not simply process information but actively reshape how humans think, reason, and interpret reality.

The review also places eco-cognitive computationalism in dialogue with broader movements such as Digital Humanism. While Digital Humanism focuses on aligning digital technologies with human values and social goals, Magnani’s contribution, as presented by the author, goes further by offering a philosophical explanation for why cognition itself cannot be separated from its material and cultural scaffolds. Computation, in this account, is inseparable from human life because it grows out of human practices of representation, experimentation, and interpretation.

Embodiment, abduction, and the limits of AI autonomy

A major focus of the review is Magnani’s emphasis on embodiment and abductive reasoning as defining features of human cognition. The author outlines how the book draws on established theories of distributed cognition and the extended mind to argue that thinking does not reside solely in the brain. Instead, cognition extends into bodies, tools, environments, and engineered artefacts.

The discussion of morphological computation plays a central role here. Magnani uses examples from robotics and embodied artificial intelligence to show how physical form and material structure can actively contribute to cognitive performance. In these systems, intelligence is not only encoded in software but distributed across hardware, morphology, and environmental interaction. The author notes that this approach reinforces the idea that cognition is relational, emerging through interaction rather than internal calculation alone.

Another key contribution highlighted in the review is the centrality of abduction. Abductive reasoning is presented as the generative logic that allows humans to form hypotheses, reinterpret situations, and navigate open-ended environments. Magnani distinguishes between locked strategies and unlocked strategies to clarify the limits of current AI systems. Locked strategies operate within closed, rule-bound domains where goals and constraints are clearly defined. Many high-performing AI systems fall into this category, excelling at optimization within fixed frameworks.

Unlocked strategies, by contrast, characterize human reasoning in real-world contexts. These strategies remain open to reinterpretation, context shifts, and historical change. The author explains that Magnani uses this distinction to show why even advanced AI systems remain dependent on human framing and guidance. While systems like game-playing algorithms can explore vast search spaces, they do so within boundaries set by human designers and objectives.

The review also addresses the growing relevance of generative AI. The author notes that recent models introduce forms of synthetic abduction by generating plausible hypotheses and interpretations. However, these systems still operate within human abductive ecologies rather than independently from them. Their outputs gain meaning only through human interpretation, evaluation, and integration into broader cognitive practices.

By foregrounding abduction and embodiment, eco-cognitive computationalism challenges narratives that portray artificial intelligence as an emerging autonomous intelligence. Instead, the framework emphasizes continuity between human reasoning and technological artefacts, while also clarifying the structural limits that prevent machines from fully replicating human cognitive openness.

Ethical stakes and the future of human-centered AI

As computational systems become increasingly pervasive, the way humans design and integrate these systems shapes not only technological outcomes but also the horizons of human cognition itself. The author highlights Magnani’s concern that over-optimized computational environments risk narrowing cognitive diversity and suppressing uncertainty.

A central ethical concept in the framework is the protection of what Magnani calls ignorant entities. Rather than treating ignorance as a flaw to be eliminated, the framework views it as a necessary condition for learning, creativity, and abductive reasoning. Preserving spaces of uncertainty allows humans to remain open to reinterpretation and discovery, rather than becoming constrained by rigid computational logics.

This ethical stance has direct implications for AI governance and design. According to the review, eco-cognitive computationalism calls for technologies that support human interpretive freedom rather than replace it. Systems should be designed to augment human reasoning, not to close off alternative perspectives or enforce narrow definitions of efficiency and success.

The author also positions Magnani’s approach in relation to other major theories of cognition, including biologically grounded frameworks that define cognition as a property of living systems. While these approaches emphasize the continuity between cognition and life, eco-cognitive computationalism focuses more narrowly on human cultural and technological ecologies. The review identifies this as both a strength and a limitation. On one hand, Magnani offers a detailed account of how humans integrate tools and symbols into meaning-making practices. On the other hand, the framework leaves open questions about how non-human cognition and ecological systems fit into the ethical picture.