Why Do We Pursue Science?
A Reflection on the Scientific Endeavour
My life is very, very unusual.
I spend my time attempting to construct a mathematical structure which can reason generally. To do so, I need to examine what intelligence itself is, abstracting away from any particular local or general realization.
Conceptually, this is very far removed from finding berries and hunting gazelle, the behavioural envelope against which humans are optimized. Science, generally, is far removed from direct importance.
Why do we pursue science? Why do we examine the zeros of functions, the behaviour of enzymes, and the motion of particles? I do not ask this question while hoping for a vaguely self-congratulatory answer, e.g. that we pursue science to satisfy a sense of curiosity inherent to humans. Really, why do we pursue science? At the fundamental level, science is pursued because its results are useful. The criteria of usefulness may change, but this structure is steady.
Science was first used because its results were theologically useful. Religious dynamics turned on the visible behaviour of celestial objects, so improvements in the ability to predict these behaviours ex ante improved the ability to predict religious activities. Because the complexity of celestial behaviour was sufficient that studying it through its utility to religion overly dragged progress, astronomers emerged who studied celestial behaviour directly without individually considering downstream theological utility of the knowledge they produce. Consequently, astronomy was the first science to emerge in the sense of being the first field which saw intergenerationally-sustained growth in an understanding of a matter which was not itself applied. We can contrast this with, for example, improvements to agricultural techniques in the millennia preceding the third-millennium BCE emergence of astronomy; although intergenerational growth was certainly present, it was on an applied technique, not on a body of knowledge whose value derived from having that knowledge itself.
Astronomy in this context was funded in the hopes that its findings would eventually be theologically useful, but individual astronomers and individual findings were not evaluated by directly forecasting their eventual theological utility. Instead, they were evaluated on the extent to which they advanced astronomical knowledge itself, and the eventual theological utility of this knowledge and its producers were considered separately by examining the extent to which astronomical advancement imparted theological utility.
This structure has persisted since its first emergence of science in the third millennium BCE, and it emerged in all other contexts where science emerged and lasted: individual scientists and individual scientific findings were evaluated on the extent to which they advanced science itself while scientific fields as a whole were evaluated on the extent to which they were useful for matters important. This was necessary: as bodies of scientific knowledge grow, it becomes increasingly difficult to evaluate how useful individual findings are for downstream uses, forcing an evaluative decoupling.
This decoupling is where I want to focus my attention, and it lets us sharpen our earlier question. Replacing our earlier query of why science is pursued, we can ask whether advances in science should be evaluated exclusively on the extent to which they advance their respective fields, evaluated exclusively on the extent to which they impact downstream utility, or something else.
This is a difficult question to answer. When scientific funding was limited, the difficulty of the question was immaterial as the answer to the question itself had minimal society-wide impact. However, with the rapid increase in scientific funding with the transition to the modern era, this question became increasingly salient as states and societies became compelled to justify their increasingly large expenditure on basic scientific research.
The scientific community in the nineteenth century attempted to evade this question by constructing an understanding of how basic scientific work is eventually translated into practical utility. Explicit attention was systematically paid to the chain through which findings in a given field would eventually lead to utility inherently useful, arranged into what were networks in substances even if not explicitly articulated as such. For example, it became recognized that discoveries in how specific chemical reactions proceed could plausibly lead to improvements in the collective understanding of certain biological processes, whose improved understanding could itself plausibly lead to improved medicine. As such, it became structurally routine and central to aim to articulate the pathway by which basic research could be eventually translated into practical utility.
These translational pathways have grown in scope and nuance, and for many areas of science, they obviate the need to ask the question. If a field is sufficiently mature that the form of its findings can be reliably predicted ex ante, then it is possible to identify translational pathways for its findings, obviating the need for the field to answer this question for findings which have an identified translational pathway. This is not an obscure observation—far from it. In many fields, identifying such translation pathways to avoid needing to answer this question is a major research activity, and engaging in it is not seen as defensive justification but as responsible consideration of downstream utility. Given the difficulty of this question and the possible benefits of certain scientific advances, this is less a cowardly evasion and more a case of individuals needing to conduct important work before philosophers finish answering a question.
This question does not need to be answered given the construction of an appropriate translational pathway.
My own research has such a translation pathway. Advances in our understanding of the nature of intelligence improve our ability to construct an artificial intelligence which can answer questions we ask and conduct intellectual labour we consider useful. The translational pathway is so simple as to be nearly comical: as long as there are questions and problems worth answering and solving, respectively, advances in intelligence are useful.
My expectation is that, over a very long horizon, continued economic growth will lead to this question becoming increasingly less important. If increases in scientific funding at the onset of the modern era forced this question, increases in the economic base freeing more funding for science would lessen its significance. Perhaps humanity in the distant future will look back at our current attempts to justify science as quaint endeavours when scientific funding was limited and in need of rationing.
My education was in mathematics with my education in artificial intelligence fitted atop. As the reader can expect, I deeply internalized the prevailing norm in mathematics that advances in the field should be evaluated on their respective contributions to mathematics, not on their practical utility. More than most fields, mathematics is structurally forced to address this question as its translational pathways are arguably the most complex of any field in the sciences. Its continued existence is not the result of articulations of translational pathways but through an exceptional risk–reward structure: even if almost all mathematical findings are not useful, the few which are justify the field.
I will most likely continue thinking about this question for the remainder of my life, assuming the philosophers of science do not provide an unexpected answer, even if its answer is immaterial for my research. If there does exist a satisfying answer to this question, my research in intelligence may even lead to the production of an artificial intelligence which answers it for us. My hope is that a solution is in our stars.