The emergence of Artificial Intelligence (AI) has been marked by significant historical milestones that have shaped the way we perceive and interact with technology. One of the pivotal moments in the early development of AI occurred in 1950 with the publication of Alan Turing’s paper “Computing Machinery and Intelligence.” In this landmark work, Turing articulated in academic terms the concept of a machine that could simulate human intelligence. He proposed what is now known as the Turing Test, a measure of a machine’s ability to mimic intelligent behavior indistinguishable from that of a human5. This paper laid part of the modern philosophical foundation for AI studies and development by raising questions about the nature of human thought and the potential for machines to replicate it, setting the stage for subsequent research and exploration.

Another crucial moment in the history of AI was the 1956 Dartmouth College Conference, often referred to as the birth of AI as a formal discipline. At this conference, pioneers in the field, including John McCarthy and Marvin Minsky, gathered to discuss the possibility of creating machines that could perform tasks that require human intelligence and rationality. One year before, in a formal proposal for the conference presented to Dartmouth College, McCarthy, Minsky, and others stated that they were building upon “the conjecture that every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it”6. They believed that AI could solve complex problems, such as language translation and pattern recognition, and they set ambitious goals for the field. While the optimism at Dartmouth was high, the conference marked the beginning of decades of research and development to realize the potential of AI.

Fast forward to 2022, when the global release of ChatGPT consolidated another significant milestone in the history of AI. ChatGPT is a product of the deep learning revolution in AI development and a testament to the incredible advancements in natural language processing by algorithms. With its release, AI took a giant leap in its ability to behave as it “understands” natural language and, therefore, generate human-like text, facilitating human-computer interactions at an unprecedented level. This moment underscores the rapid – and at times unchecked – development of AI and its potential to grasp a significant part of the complexity of human language from a computational standpoint. It also raises important ethical and societal questions about the responsible use of AI and its implications for various fields, from elections to employment.

Against this significant (albeit historically recent) backdrop, an important conversation on AI regulation gained traction. See, for example, the AI Act that was approved by the European Union Parliament in 2023 and is now undergoing a final round of negotiations. The regulation focuses on possible negative externalities of the technology and places AI on a spectrum ranging from low risk to unacceptable risk. This is set to become a very influential framework worldwide, following the now well-known “Brussels Effect”7 that can also be identified in the field of privacy & data protection and platform regulation. See, for example, Brazil’s Draft Bill n. 2.338/2023 that was heavily influenced by the European debate and also follows a risk-based approach instead of proposing a regulation of the technology itself. Moreover, the United States also joined the conversation in 2023 when President Biden issued a breakthrough “Administrative Order on Safe, Secure, and Trustworthy Artificial Intelligence”, paving the way towards a more balanced framework for AI development and application.

Nonetheless, the concept of technodiversity challenges some of the preconceived notions on which these conversations are grounded. As Yuk Hui puts it, we should “liberate machine intelligence from the bias of certain notions of intelligence” so we can be free “to conceive new political ecologies and political economies of machine intelligence”8. To the author, the prevailing account of AI development – which was briefly presented above – casts a shadow on the “epistemological rupture” between a linear form of reasoning and recursivity, which lies at the foundation of new theories such as cybernetics and systems theory. Hui argues that “the recursive form allows the algorithm to effectively absorb contingency to improve computational efficiency”9. In other words, a recursive form of reasoning allows the machine to derive its rules from experience instead of depending solely on the rules set by its programmer. Modern AI-based technologies are capable, therefore, of non-linear reasoning, which differentiates them from a classic mechanistic thinking and moves them closer to the behavior of organisms. Exceeding the “soulless automaton” denounced by Renè Descarters, “the circular causality [...] seems to suggest a movement analogical to the soul: the soul is that which returns to itself in order to determine itself”10.

​

Even though the development of AI has moved in the past decades from a strict Cartesian rationalism (as philosopher Hubert Dreyfus has been ceaselessly criticizing since the 1960s) towards an empiricism that connects decisions to fact-based analysis, of which connectionism has been a paradigm, it undermines the diversity of intelligences and their complex relations to the world.11 In Hui’s words, paraphrasing Heidegger, “the world is constituted by a complex totality of references, and cognition depends on these references in order to reason”12. Despite observing the increasing power of AI, one should not lose sight of how the world itself is changing, moving away from a rich and complex phenomenological reality, due to an accelerating process of digitization, to a world increasingly determined by data and mathematical calculations. We are talking, therefore, of an intelligence that is computable (or recursively enumerable), which, as Hui aptly acknowledges, “is only one type of intelligence among many others”13.

The very concept of technodiversity, as mentioned earlier, is intimately related to the concept of noodiversity, which refers to the diversity of reasoning or thinking. It is, therefore, a call to problematize and reconceptualize the very idea of intelligence baked into the field of Artificial Intelligence as a whole. See, for instance, how Chinese philosophy attributes intelligence to an intellectual intuition while western philosophers tend to reject this affirmation14. Yuk Hui, in light of this analysis, suggests that we should enlarge our understanding of intelligence in two ways in order to, consequently, enlarge our understanding of how technology works. First, intelligence should not be limited to calculability, as it is now considered and implied in the competition of computational power. Instead, it should take the incalculable, which underlies in all spiritual lives, as its departing point. Second, intelligence should not be reduced to a homogenous universal model which currently leads to the search for a universal super-intelligence. Instead, as Hui puts it, AI should be informed by the goal of facilitating and not hindering noodiversity, which, in turn, fosters technodiversity and biodiversity15.

All in all, this means that we should not focus on the pursuit of a universal AI driven by speed and efficiency. Instead, AI needs to be assessed from the point of view of an enlarged conception of intelligence alongside various and diverse modes of reasoning or thinking (noodiversity). It is this recognition that will lead us to consider alternative paths and, moreover, meditate on the possibilities of AI beyond a universal super-intelligence (or on how to prevent this illusion from driving us towards the abyss). This should include a discussion on the diversity of intelligences, the development of algorithms that power AI-based solutions and the implementation and application of AI tools under diverse circumstances.

This universalist undertaking distracts us from a more fruitful debate on how AI can advance, instead of hinder, technodiversity and how this, in turn, can have a positive impact on the development of noodiversity and biodiversity. Hence, regulators and policymakers around the world must consider technodiversity as an element when drafting and implementing new normative frameworks for AI, contributing to the creation of a more diverse (instead of monolithic) technological future. This means exploring other paths beyond the prevailing narrative, which is built on an universal (and, therefore, questionable) approach to technology.