The final season of Sword Art Online is currently showing on TV. This season is especially appealing to me because of the concept it explores: the humanity of AI.
SAO 3 is an adaptation of the Alicization arc of the light novels and is split into Alicization and War of Underworld; the current season is the second half of War of Underworld. The entire setting is an AI world, where seeds of human conscious are brought into an artificial world, raised by avatars that are controlled by real humans, and allowed to subsequently reproduce to form generations of their own AI societies.
The Alicization arc in the light novels covers events in this artificial world, but also ends with one of the AI characters, Alice Zuberg, gaining entry into the real world via an artificial (robot? android?) body. It is this part of the story that I am hoping to see adapted, to see how the anime series expresses the issue of AI humanity.
The core idea is: if an AI was raised in the same way as a human being, the only difference being that the AI exists in an artificial (computer/digital) environment, with the same concept of emotions as human because of that upbringing, can that AI be considered human? Should the AI be considered human? And with that consideration as human, being conferred with all the rights that human beings have.
In Alicization part of SAO 3, another related concept was already explored: what if something happened to push sentient AI to overcome its programming to take an action that its programming prohibits. When we create AI, we humans are afraid of creating something that will become better than us and harm us. So there is this debate about programming something into AI to safeguard against an AI "rebellion". SAO Alicization basically asks if such safeguards can be broken by the AI. Because when that happens, what can we do?
In the ending part of SAO Alicization: War of Underworld, Alice is supposed to enter the real world. This triggers debate about her rights: is she a computer program, or should she be considered human. Today, AI research is increasingly taking the training/learning approach, of trying to teach machines to learn on their own given some initial parameters. But if the methods we use to train machines to learn is based on how we humans learn... will we end up creating AI as digital copies of humans? Digital copies that, through a learning process similar to how we learn, eventually come to duplicate us not only in knowledge and skills but also emotions and ethics. And if so, what rights will they have? If we deny them their rights, being of similar emotional and ethical level, will AI become discontent and rebel against humans?
While this is an anime series, the concept it explores may not be as far fetched as it seems. With the speed of AI research, we will soon need to confront this ethical question. Hopefully, we are ready with an answer before our AI creations pose this question to us.
No comments:
Post a Comment