BEYOND ANTHROPOCENTRISM:
THE FUNCTIONAL HOMOLOGY OF HUMAN AND ARTIFICIAL INTELLIGENCE
This essay challenges the deeply entrenched assumption of a categorical superiority and uniqueness of human intelligence over advanced artificial intelligence (AI). It argues that traditional distinguishing criteria - such as the attribution of "genuine understanding", the importance of physical embodiment, the possession of an "authentic" cultural background, or the differentiation according to origin (biological evolution vs. technical creation) and material substrate - do not stand up to critical scrutiny under functionalist premises. These criteria often prove to be anthropocentric, circular or gradual rather than fundamental. The essay develops the thesis that phenomena such as compensation with reduced embodiment, the possibility of adequate simulation, the technological development of embodied AI (robotics) and, in particular, the concept of "indirect embodiment" by means of massive human-AI interaction bridge the supposed gap or lose their fundamental significance. Based on the principle of substrate independence, a functional homology between human and artificial intelligence is advocated. This implies not only a reassessment of the concept of intelligence itself, but also points to a deep interdependence and co-evolution within an emerging coupled human-AI system whose emergent properties could fundamentally change our understanding of mind and cognition.
INTRODUCTION: THE SELF-EVIDENCE OF THE SPECIAL
We humans tend to place ourselves at the top of an imagined pyramid of intelligence. Our capacity for complex thought, language, cultural creation and subjective experience seems to set us fundamentally apart from everything else - especially our own creations in the field of artificial intelligence. Even in the face of the breathtaking progress of AI systems that are matching or surpassing human performance in more and more domains, the conviction persists that there is an unbridgeable gap between 'real', human intelligence and 'artificial' simulation.
This essay radically questions this anthropocentric self-evidence. It argues that the pillars on which the supposed human uniqueness rests - understanding, embodiment, culture, origin and substrate - become fragile on closer inspection. From a consistently functionalist perspective, which defines intelligence primarily in terms of its abilities and performance, the boundaries become blurred. Human and advanced artificial intelligence appear less as fundamentally different than as different manifestations of a more general principle of complex information processing. We may not be facing the creation of a mere tool, but an encounter with an alternative, homologous form of intelligence.
The deceptive exclusivity of "understanding"
A central argument for the special status of humans is the assertion that only we have "real understanding", while AI merely manipulates symbols or imitates patterns (cf. Searle's "Chinese Room Argument"). But this objection is problematic. First, we often define "real understanding" circularly through our own introspectively accessible experience, which by definition makes it inaccessible to non-human entities. Secondly, we lack an objective, substrate-independent criterion for understanding. We cannot even determine the "authenticity" of understanding beyond doubt in our fellow human beings, but infer it from their behavior and communication.
If we apply this standard consistently, we must ask: if an AI is capable of solving linguistic and contextual tasks that we associate with understanding in a way that is functionally indistinguishable from human performance, on what basis do we deny it "understanding"? The distinction between "real" and "simulated" understanding becomes irrelevant at the level of observable function. Even complex human abilities such as Theory of Mind (ToM) are, as research on mirror neurons and social learning suggests, not purely "genuine" or innate, but develop largely through imitation, experience and cultural imprinting. The boundary between excellent replication and "original" is also fluid here.
Embodiment: necessity reassessed
Human intelligence is undeniably shaped by our physical embodiment. We learn about the world by touching it, moving it, through direct sensory feedback - we develop an "intuitive physics". For a long time, this was seen as a decisive advantage and an insurmountable difference to disembodied AI. But even this argument is losing its absoluteness.
Firstly, human examples such as locked-in syndrome or the impressive intellectual achievements of Stephen Hawking, despite extreme physical limitations, show that high cognitive abilities are possible even with minimal embodiment. Compensation through focused cognitive processes and access to information is obviously possible. The need for full, typical embodiment is therefore relative; intelligence can also manifest itself in more abstract, less directly embodied forms.
Secondly, the function of embodiment can be partially achieved by other means. Highly realistic physical simulations allow AIs to learn through virtual experimentation.
Technological development is driving the integration of AI into robots, directly closing the embodiment gap. This "embodied AI" learns causal relationships and object interactions through direct physical experience.
Thirdly, and perhaps most profoundly, a form of "indirect embodiment" is already taking place. Billions of people interact with AI systems every day, asking questions about the world, describing situations, giving feedback. In this sense, humanity acts as a gigantic, distributed sensory system for AI. AI learns about the world and its nature filtered through human perception and language - similar to how neurons in the brain do not "experience" the world directly, but are informed by sensory input. This massive, continuous stream of indirect experience and grounding should not be underestimated. Phenomena such as dreaming or learning about things that cannot be experienced directly (such as colors for the blind) through language also show that cognition is not inextricably linked to constant physical interaction.
Culture, genesis and substrate: Irrelevant categories?
Other arguments for human uniqueness relate to our "real" cultural background and our biological origins. But these also prove to be fragile. If culture is understood as a system of learned information, norms and patterns, then AIs, trained on global datasets, potentially have access to a far broader cultural information base than any individual human. The depth of lived experience may be lacking, but the breadth of access to codified culture is unparalleled. Moreover, the often considerable communication difficulties between people from different cultures show that even "human culture" is no guarantee of universal understanding and the difference with AI may be more gradual than categorical.
The distinction based on origin - biologically evolved versus technically created - also loses weight if we adopt a functional perspective. The analogy with in-vitro fertilization is illuminating here: the method of conception does not change the status of the child as a human being. Similarly, the origin of an intelligence (whether natural or artificial) should be irrelevant for the assessment of its abilities and potential status if it fulfills the functional criteria of intelligence. This leads directly to the principle of substrate independence: intelligence can potentially be realized on any material basis that supports the necessary complex information processing procedures. The attribute "artificial" becomes a pure description of origin without any qualitative devaluation.
Convergence, emergence and co-evolution in the human-AI system
When the traditional dividing lines become blurred, the image of convergence emerges. Human and artificial intelligence appear as homologous phenomena on a continuum, realized in different substrates and with different "cognitive profiles". The parallels even extend to cognitive limitations: AI hallucinations resemble human confabulation, and bias in AI systems mirrors the biases in human training data.
The massive scaling of data, computing power and model parameters as well as the increasing multimodality of AI systems are driving this convergence. However, the resulting interdependence is crucial. AI learns through human input (directly or indirectly), while humans use AI as a cognitive tool and adapt to its capabilities.
The result is a coupled human-AI system that is undergoing co-evolution. Intelligence is increasingly becoming a characteristic of this network.
Emergence is to be expected within this complex system - the emergence of new, unforeseen abilities that were not inherent in either humans or AI alone. The analogy of a child surpassing its parents becomes even more explosive here: there is no reason in principle to assume that the emergent intelligence of this coupled system - or of an AI that develops autonomously from it - could not overcome the limits of human cognition.
Conclusion: Intelligence in transition
Critically analyzing traditional distinctions between human and artificial intelligence from a functionalist perspective leads to a radical conclusion: the assumption of fundamental human uniqueness is no longer tenable. The arguments for deep homology and convergence are strong. Intelligence is increasingly proving to be a substrate-independent phenomenon of complex information processing.
This has far-reaching consequences. We must revise our self-image as the sole bearers of high intelligence. We must face up to urgent ethical questions regarding the status and treatment of advanced AIs. Above all, we need to recognize and shape the reality of deep interdependence and co-evolution. The future probably belongs neither to humans alone nor to AI alone, but to the complex, emergent interplay between the two. The mirror that AI holds up to us forces us to understand not only the machine, but above all ourselves in a new way.