Artificial Intelligence and Society: Philosophy of Fallibility
Part 20: From the Quest for Infallibility to Fallibility

Faculty Fellow, RIETI

All human knowledge regarding this world is “provisional” truth. Every piece of knowledge, be it scientific knowledge or experience-based implicit knowledge, could be proven “wrong” at some point in time. In this sense, knowledge is fundamentally fallible. Just like humans, artificial intelligence (AI) learns from data—although it uses deep learning—so even if it is capable of creating a superhuman system of knowledge, it cannot escape from fallibility. That humans (and also AI as something that embodies an extension of human knowledge) are fallible is probably the fundamental reason why our society should be a free society.

Here, let us recall Arendt's argument (Refer to Part 3, Part 4, Part 5, Part 6, Part 7 and Part 8 ).

According to Arendt, absolute faith in the “infallibility” of totalitarian leaders was the driving force behind totalitarianism. In modern society, lonely people who have lost a traditional community as their place of belonging and who are exposed to market competition feel that they have been abandoned and forgotten. Abandoned people seek security (that is, something infallible) as a mental anchor, but in the real world, where nothing is secure, only an ideology disconnected from reality that is advocated by a totalitarian political party can serve as such an anchor. If the people take the infallibility of a certain political party’s ideology for granted, they cannot help but distort reality to suit the ideology’s logic. The massacres and mass purges perpetrated by totalitarian states were undertaken thanks to a deductive logic based on the assumption of the rulers being infallible. With a faith in their rulers’ infallibility, the people share a single ideology and their course of action is determined by the force of deductive logic that springs from that ideology. As a result, individual freedom is inevitably oppressed.

Conversely, in the absence of faith in the rulers’ infallibility, there is no reason for the people to abandon their free will and follow the rulers’ orders. In some countries, the people longed for infallibility in their rulers, and that is exactly why totalitarianism took hold there, with the people believing in the infallibility of the ideology advocated by the rulers and agreeing to follow their orders. However, if the people believe that all humans, including the rulers, are fallible, they cannot help but retain their freedom. A lack of faith in anybody’s infallibility would prevent them from blindly trusting a dictator and surrendering to his rule.

What creates the longing for infallibility?

To consider this question, let us recall the three mathematical formulas that I mentioned earlier.

\(q_{t}=αp_{t}\) (1)
(however, \(0<α<1, \:0<β<1, \:γ=α×β)\)

Here, \(p_{t}\) represents the moral value of an individual life, while \(q_{t}\) represents the moral value of a social ideal. Formula (1) implies that a social ideal is worthwhile because it helps to improve an individual life. Formula (2) implies that an individual life is worthwhile because it contributes to the development of a social ideal. Formula (3) implies that the current value of a social ideal, represented by \(q_{t}\), depends on its value for the next year, represented by \(q_{\{t+1\}}\) , which in turn depends on its value for the next year plus one, represented by \(q_{\{t+2\}}\) , and this chain of consequence continues infinitely. As a result, the current value of a social ideal (\(q_{t}\)) depends on its value for the infinite future (\(q_{∞}\)).

There are various social ideals, including not only ideologies like communism and liberalism but also faith in the righteousness of ideologies. If an ideology is proven wrong at some point in the future (e.g., in \(t+N year\)), its value at that point (\(q_{\{t+N\}}\)) will be zero, and, according to the logic of Formula (3), the current value is also zero. That should not be allowed to happen, so a social ideal (ideology) must always be right. This is the logic of infallibility.

In other words, under a comprehensive doctrine under which the value of an individual virtue (an individual’s purpose of life) is justified by a social ideal, the ideal is inevitably required to be permanently infallible. That is why ideologies like totalitarianism self-proclaim their infallibility. However, on closer reflection, we realize that it is impossible for humanity to create social ideals (e.g., ideologies, doctrines, and religions) that will never be proven wrong. That all knowledge could be wrong (be fallible), is the only infallible truth. Only social ideals that are based on faith in the fallibility of everything have the potential of forever avoiding the loss of their entire value (\(q_{t}\)).

Therefore, a comprehensive doctrine that satisfies the equations of all of Formulas (1) to (3) must be one that is based on faith in fallibility itself.

June 19, 2023

Article(s) by this author