Knowing your robot: the fiduciary program in the age of AI | The Polanyi Society

Select Page

Knowing your robot: the fiduciary program in the age of AI

Mihály Héder

To a longtime supporter of our department in Budapest,

to a friend,

to Phil Mullins.

Introduction

Michael Polanyi is one of the most influential philosophers of science of the twentieth century, although perhaps he is not as well known as he is important. While his older brother, Karl, made a highly impactful, well-known contribution to the understanding of macroeconomics in his seminal work The Great Transformation ([1944] 1957), Michael Polanyi made a similarly impactful contribution to, well, all areas of thinking, really, with Personal Knowledge: Towards a Post-Critical Philosophy (1958).

This contribution, besides being nutritious food for thought for philosophers, also influenced the everyday work of engineers, designers, and managers, even if only indirectly. Sometimes we can even find cases of direct influence, as in the case of Xerox Parc.1 The concept of tacit knowledge appears in the everyday considerations of engineering and bureaucracy and is cited almost as often today as in any of the other five decades that have passed since publication.

One wonders if the continued relevance of post-critical philosophy, which is apparent in the citations, is because Michael Polanyi’s central message is still unheeded; the perception of a person’s epistemic faculties and ontological status is still as distorted, at least in the grand scheme of things, as it was in the late 1950s, to which Polanyi responded. I will argue that this is indeed the case: the same self-defeating variant of critical approach Polanyi tries to surpass is still prevalent in the contemporary debates. One such debate is about the epistemic faculties and the moral and ontological status of robots, that is, the machine question (Gunkel 2012).

As to the source of this framing problem, the clue is in Michael Polanyi’s Personal Knowledge subtitle: the overly zealous critical philosophy.

The entire post-critical project is often seen by many as a major contribution mostly to epistemology, but I argue that, while this is true, epistemology is a means to an end; it is applied to issues about organization of society and, ultimately, to profound moral questions, which are all part of Polanyi’s message.

Post-critical philosophy

There is no better source for understanding the evolution of Polanyi’s thinking on the matter than Phil Mullins (2001). He explains the origin of the term "post-critical philosophy" and, more widely, the vision Polanyi offers. Mullins traces the term back to 1951, to the first series of the so-called Gifford Lectures. At the center of it is an emphasis on the indispensable personal participation of the knower in the known.

I will argue in this paper that research on our personal involvement in evaluating artificial intelligence (AI) will prove highly fruitful and insightful. Mullins positions Polanyi as a thinker who was trying to process the implications of the findings of Gestalt psychology, in that the whole perception cannot be reduced to the sum of its parts. I intend to frame the evaluation of AI with this in mind.

We also learn that by "critical" Polanyi means "Descartes, Hume, Kant, J. S. Mill, and Bertrand Russell" and the great success their systematic doubt brought to science. However, he also believes that this approach is reaching its limits as a heuristic, and a new perception is required about the methodology of scientific thinking. This is post-critical philosophy, which, in terms of history of science, can be called postmodern thought as well. However, as Mullins cites several sources on the matter, postmodern in this case is quite constructive and commitment-oriented, contrary to the usual associations of the term. But post-critical thinking tries to supersede not only unbounded doubt as the central heuristic of science but also the ideal of impersonal knowledge. To do that, Mullins explains, Polanyi rehabilitates confidence in "overt belief" (Mullins 2001, 78).

Together with the active involvement of the person, and the role of commitment, post-critical thinking also opposes all kinds of dualism without itself becoming monistic; through the theory of emergence, it rejects the dichotomy, and therefore the dilemma, between monistic and dualistic worldviews, and Mullins guides us to the works of Sanders (1991) and Gill (2000) on the matter.

The stakes

Before we can arrive at the question of AI, we need to understand just how high Polanyi’s aspirations were when he wrote Personal Knowledge. Starting from understanding an epistemic-ontological issue, he reframes the history of science, not only for deepening our knowledge of past events but also to reestablish the methodology of science. In his historical approach he even precedes Thomas Kuhn. At the same time, he derives some prescriptions, not only for science narrowly but...