Emanuele Martinelli

Digitalization and AI are bringing about the automation of many aspects of our market economies. We shall argue, however, that there is one market phenomenon which AI will never be in a position to automate, namely the phenomenon of entrepreneurship. Entrepreneurship is in several respects uniquely human. It involves on the one hand creativity, which is the ability to produce something that is novel but at the same time salient to outsiders. And on the other hand it involves the feature of alertness, that is the ability to detect unpredictable features of the surroundings of the entrepreneur. Humans are uniquely aware of their surroundings because they are able to draw on ways of gaining knowledge about such surroundings in virtue of their existence as living human organisms. These ways of gaining knowledge are often unconscious, and the knowledge itself provides immediate links to avenues of future response. AI systems, on the other hand, can only handle mathematical algorithms created on the basis of historical training data. We will make the claim that humans can deal with novel situations in creative ways because, unlike AI systems, they can develop the necessary practical know-how through lived experience of the relevant market phenomena. We will then show that even the most powerful generative AI models that we can foresee cannot harness practical knowledge and therefore will fail to emulate human alertness. Our argument has implications not only for the understanding of entrepreneurship but also for the so-called calculation debate. If we are right, then digitalization and AI will never change the fact that human entrepreneurs situated in market contexts are irreplaceable components of the market economy. The argument thereby implies a critique of attempts to replace individual human entrepreneurs with collectives. Neither AI models nor social organizations can achieve anything like 'cooperative entrepreneurship' or 'intrapreneurship' in a way that would diminish the essential market role of the individual human entrepreneur.

The philosophy of agency has accomplishised a good deal of work on the notion of agency. In mostly focusing on deliberated, human agency, though, little has been done to understand agency as a general phenomenon, beyond the human: what kinds of agents are there and what kinds of things may qualify as agents? In this spirit, I will advance a general taxonomy of agents, thereby including fully-formed rational human agents all the way to more thermostats or riots. I will use this conceptual framework to argue that biological agency is intrinsically superior to artificial agency, both collective and automated. In Section 2, I will introduce a notion of agency based on a development of Searle's theory of intentionality. In Section 3, I will define three levels of agency: active, reactive, and proactive; based on that, I will present a taxonomy of agents with three kinds of agents – natural, collective, automated – classified through varying degrees of agency. Finally, in Section 5, I will discuss the limits of non–human agency, precisely arguing against the existence of collective and artificial persons. If granted, this has important implications in that human agency would enjoy systematically superior agential dispositions than artificial agents.

Mises’ and Hayek’s arguments against central economic planning have long been taken as definitive proof that a centrally planned economy managed by the government would be impossible. Today, however, the exponential rise in the capacities of AI has opened up the possibility that supercomputers could have what it takes to plan the national economy. The ‘economic calculation debate’ has thus reignited. Arguably, this is because neither Mises nor Hayek have given a clear and conclusive argument why central planning of the economy is impossible in principle. The paper frames the problem of economic planning as an agent–environment interaction, offering a taxonomy of the different sets of agents at play a) in a market economy and b) in a centrally planned economy equipped with the most sophisticated AI technology. The argument is that public institutions as planning bodies cannot replace the market order, no matter the AI technology behind them, for the elimination of the market entails the elimination of crucial kinds of agents that cannot be recreated or emulated through AI or careful social planning: the proactive action of entrepreneurs driving market allocation.

Landgrebe and Smith (2022) have recently offered an important mathematical argument against the possibility of Artificial General Intelligence (AGI): human intelligence is a complex system; complex systems have some properties that cannot be modelled mathematically; hence we have no viable way to build an AI that would be able to emulate human intelligence. The issue of complexity is thus at the heart of the Landgrebe and Smith approach, and they tackle this issue by postulating a set of conditions, derived from mathematics, for a system to be complex. I argue that Hayek’s “Theory of Complex Phenomena” offers an interesting alternative path to understanding what makes complex systems complex. This offers an argument complementary to that of Landgrebe and Smith, which approaches the issue of complexity from an ontological angle, and which is designed to show that AI systems are not the right kind of entity that may interact meaningfully with complex systems.

Our goal in this paper is to establish a set of criteria for understanding the meaning and sources of attributing (un)fairness to AI algorithms. To do so, we first establish that (un)fairness, like other normative notions, can be understood in a proper primary sense and in secondary senses derived by analogy. We argue that AI algorithms cannot be said to be (un)fair in the proper sense due to a set of criteria related to normativity and agency. However, we demonstrate how and why AI algorithms can be qualified as (un)fair by analogy and explore the sources of this (un)fairness and the associated problems of responsibility assignment. We conclude that more user-driven AI approaches could alleviate some of these difficulties.

Generally speaking, the present literature on agency has been heavily focused on human agency. This approach proves to be very useful for the immediate applications of the philosophy of agency, e.g. to develop a definition ready for use in ethics or political philosophy. However, there are some limitations to this line of thought, as, for instance, it poses too restrictive requirements on agency, like purposefulness, consciousness, or willingness. In this paper, I would like to tackle the question of agency with the need to include non-human agency in mind. I will set the foundations for a conceptual framework to define agency in the most general terms possible. This framework should be able to capture the conditions at which something possesses agency and the features of different kinds of agents in a way that includes animals and plants, mobs, institutions, ecosystems, AI machines, artifacts, and so on. In general, I claim that agency is based on a weak notion of intentionality, ‘directedness’, which does not require mentality or purposefulness.

_Posthuman Bliss? The Failed Promise of Transhumanism_ by Susan B. Levin presents a well-informed and structured critique to transhumanism. Not only transhumanist ethical and sociopo-litical applications are challenged: the theoretical assumptions and implications of transhumanism are made explicit and put into discussion, thereby confronting transhumanism as a worldview of its own, from metaphysics, to epistemology, to philosophy of mind, down to ethics and politics. This worldview is then constantly put to the test of Levin’s own Aristotelian essentialist framework, which sees the human being as a holistic whole and our role in the world as the complex process of attaining human flourishing. First, I will spend a few words to frame the general aim of the book and to delineate the main contents of the seven chapters. After that, I will deal critically with four interesting focal points, where further research may be prompted or the standpoint of the author may be challenged: the attribution of (rational) essentialism to transhumanism, the definition of well-being underlying transhumanist positions, the apparent tension between the sociopolitical im-plications of transhumanism and its cultivation of radical personal autonomy, the critical evaluation of Levin’s American-centered adaptation of her Aristotelian virtue-ethics approach.