Comparing natural and artificial agents

Between the natural and the artificial agents, there are many differences, but few similarities. An inclusive definition of agency needs to cater for all these agents, in all the diversity they exhibited in the preceding posts. At this stage it is useful to review the major differences, at least conceptually, because any definition of agency will be inclusive if —and only if— it remains agnostic of those differences across the diversity of agents. Two aspects stand out as particularly diverse: the agent’s impact and its endurance.

Impact describes to the effect an agent can cause directly on its environment. We easily recognise the high impact that each human can have, at least when compared with the low impact of a single fungus or algae. Some organizations can have super impact (think about the World Bank), while the impact of some technologies is potentially significant (consider AI again), though not yet fully known.

• However, beyond such superficial comparisons, even agents with the highest potential impact are not omnipotent. Their actions are limited by their own means (e.g., an industrial robot might have full autonomy over its motions, yet it remains bolted down in one location; it cannot move around), by their environment (for all living organisms, their respective ecological niche defines whether they will flourish or perish) or by other agents (a balance of power, e.g., between local and federal governmental responsibilities). No agent can wield unlimited power.
• On the other hand, even agents with negligible individual impact can develop decisive might. For example, the orchestration of many similar minor forces can create effects well outside the control of any of the individual contributing agents (that is the ‘logic’ of swarms and flocks, of troops and groups, and of super-organizations in general). Alternatively, different types of agents may team up and form a powerful symbiosis. Consider the mycorrhizal networks that allow fungi and plants to exchange nutrients and energy, or the seamless symbiosis between the companies and technologies that define the platform economy. Every agent can contribute to effects beyond its immediate control.

Endurance refers to the longevity of the agent, or at least of its agency. At first glance, that might seem to defy conventional wisdom, but highlighting the difference between the agent and its agency will prove instructive.

• The life-cycle of natural agents is predetermined and limited by nature. Though this life expectancy literally ranges from hours to ages, every living being will eventually die. Artificial agents, however, do not have a predefined death. Some companies might die very fast (start-ups are notorious for what their founders will consider a premature death), while states or religions may flourish and endure over many centuries (consider China or the Vatican). Technologies are yet different: because we cannot ‘un-invent’ them once the underlying ideas have been tried and tested, they cannot lose their basic effectiveness. Therefore, they have no end of utility in sight, despite potentially becoming outdated, out of use, or even forgotten.
• An agent does not need to die in order to loose —even if only temporarily— its agency. Some agency wears off by using it. Take the honeybee as a case in point (or jellyfish, or any other venomous being): the production of such venom consumes time, material, and energy; replenishing an exhausted storage will take some time during which the venom is unavailable (and the animal or plant in question is rather defenceless). A similar correlation applies to some artificial agents as well, just consider monetary policy and the stability of a currency. In this scenario, the central bank is the agent that uses interest rates as its means to keep inflation at bay. However, no central bank can lower interest rates infinitely. And if it was to adjust interest rates too frequently (either up or down), it might destroy the public’s trust in the bank and in the currency alike, which could cause inflation to spin out of control.

But what do the natural and the artificial agents have in common? What are the shared characteristics of an elephant and an ant, a giant sequoia and a stinging nettle, a fungus and an amoebae? What do they have in common with humans, with technology, with organizations? And what are the common traits shared by a citizen and her government, by the World Health organization and IBM, by ChatGTP and a newspaper? Despite this mind-blowing diversity in scale and scope, you can distil a small number of essential characteristics that all natural and artificial agents share: they have “some power” to “do something” that “somehow suits their purpose“.

Time to dare a definition of agency.

This is the seventh in a series of posts on the agency and how it matters to innovation.