Artificial Intelligence and Society: Philosophy of Fallibility
Part 22: Fallibility and Freedom

Faculty Fellow, RIETI

The way for humankind, as fallible and free beings, to make progress, is to do what could be wrong based on possibly wrong knowledge and to learn from the results. Accumulating real-world experiences of success and failure through recurring trial and error and adjusting theories based on reality is a process that is common to both human learning and AI’s deep learning. Only when the “freedom to make mistakes” is ensured is learning through that process possible.

In other words, freedom is indispensable to our society because nobody knows all the answers regarding the challenges that our society is facing and the future of our society. While innovations update the system of justice daily and promote the progress of reason, nobody knows what the ultimate destination of the progress is. All that human reason (enhanced by AI) can do is to exercise the right of freedom to engage in trial and error, acquire information on new features, and attempt to comprehend the structure of the universe through approximate calculation. However, the progress that is achieved is always open to fallibility—that is, there is always the risk that the progress may later be proved “wrong.”

While believing in the value of the progress in AI-enhanced reason and the “system of justice,” we cannot rid ourselves of doubt and trepidation over the possibility that the progress, given such fallibility, could, in fact, be wrong. Rather, it is exactly because of that kind of uncertainty that we humans do not remain content with the status quo but are willing to try something new that goes beyond convention. That is the driving force of innovation. The accumulation of innovations updates human knowledge and promotes progress in the social “system of justice.” Therefore, we may say that recognizing the fallibility of ourselves and others is the driving force of the progress of our society. As a condition for enabling individual persons to engage in trial and error and to achieve innovation, there must be a social system that ensures the first principle of Rawlsian justice, that is, basic freedom for everyone. Ensuring the freedom to engage in trial and error is a prerequisite for societal progress. In other words, ensuring freedom for individuals satisfies the condition for the moral value \(q_{t}\) in Formula (3) to turn positive and continue to increase forever through time under the moderate comprehensive doctrine that we advocate.

August 21, 2023

Article(s) by this author