Karl Popper argues that all science begins with philosophical problems and ends with philosophical problems. In the field of Computer Science, the disciplines grouped under the denomination of “artificial intelligence” (AI) do not seem to be an exception. A cursory glance at the seminal writings of the founders and the developments of pioneers in the field reveals that their original concerns are of a philosophical and epistemological order. In a second stage, the formulation of derived problems in conjunction with the search for scientific answers establishes, more or less organically, some substantive theories of AIs. This somewhat fragmentary theoretical framework –and still lacking in general consensuses– constitutes, together with the evolving operational theories, the technical scaffolding that supports AIs as formal objects and technological artifacts.
In the early decades of AI –from the 1970s to the 1990s– many in the research community broke with the earlier hermeneutics and saw reflections on AI from philosophical backgrounds as unwanted intrusions. Over time, this stance has softened, favoring instead a fruitful critical exchange. However, this dialogue has three major adversaries. First, the systematic epistemic encroachment of many developers who treat hypotheses as conclusions and make extrapolations between fields of knowledge with non-interchangeable foundations and methods. Second, the simplifications of the popularization and futurology discourse that establishes narratives of extreme optimism and apocalyptic pessimism alike. Lastly, the forces in tension between the different actors of the system –industry, academia, institutions and government–, where competition for funds and markets intersect with the discussion of ethical and regulatory aspects.
Currently, the Philosophy of Artificial Intelligence (PhilofAI) can be considered as an interdisciplinary specialization within the Philosophy of Science directly pertaining to the foundations of the Natural and Cognitive Sciences. PhilofAI addresses questions central to the definition, limits and implications of a set of Computer Science techniques that emulate the results of operations that in humans are called perception and knowledge, and which are commonly associated with intelligence. The Philosophy of Artificial Intelligence proposes the precise formulation of those questions that arise from the invention of AI: whether machines can think, learn or possess consciousness. It also aims to give them the best possible answers, articulated in terms that are consistent with current science and philosophy. In this regard, it seems pertinent to resort to Gaston Bachelard’s definition, as it spells out the very purpose of a Philosophy of AI: “science is always preceded by an epistemological obstacle that only philosophical thought can overcome”. In its theoretical approach and daily practice, AI uses terms and ideas that are borrowed from other scientific fields and based on philosophical concepts that, in many cases, are still under revision, such as intelligence, consciousness or the relationship between language and thought.
When compared to other cognitive or natural sciences, Computer Science as a whole is relatively young—those other disciplines enjoy several centuries of head start. In particular, the early years of AI, still far from reaching a century of development, reveal that despite its evolution in terms of paradigms—logical or probabilistic—methodologies—algorithmic or neural—and computational models—binary or quantum—it is still far from being considered a mature field.
In light of this short lifespan, there is, however, a long historical perspective that proves relevant to philosophical inquiry into AI: the human aspiration to construct—or (sub)create—an intelligent being, a rational “other” in humanity’s own image or even one that surpasses it. While this theme may verge on the mythical from a narrative standpoint, it nonetheless offers a legitimate avenue of exploration into the intentional origins of AI, the motivations of historical precursors and contemporary actors, and a deeper analysis of what humans seek and how they relate to the artifacts they create.
Member of:
