Will AI possess moral values?

For an author like Kant, who is a human by the way, it is easy to argue for reason and how they lead to the moral values we have. His argument about moral laws being products of rational reasoning makes sense for all of us as we have grown up being taught about some common values and certain behaviors that we should abide by. However, it is not obvious to me that there is just one kind of reasoning in this universe. I think Kant is ignoring the fact that there is a wild difference between different forms of reasoning.

An example of a wildly different form of reasoning can be realized in an artificial intelligence medium. While it is possible for us to create AI models that think like us and follow the same rules as we do, this is merely imposing our cultural biases and prejudices onto these models and making them think like us. Extensive research has shown that, in reality, we don’t understand an AI model’s logic for decision-making. When visualized, we have seen that AI models can arrive at similar conclusions as humans in certain scenarios (like diagnosing tumors from CT scans of the brain). However, they are able to do so with a significantly different methodology compared to what human doctors to by focusing on key aspects differing from those used by humans. So, if AI models’ way of reasoning is different to ours, we should also expect them to arrive at different conclusions on what’s right and what’s wrong and build different moral values.

The part about Machine Ethics is relevant in the following page:


2 thoughts on “Will AI possess moral values?

  1. Nabi, interesting topic and point! I guess I would not have thought about this, in terms of Kant not taking account for the range of reasoning that can be developed. It seems, to your point, that many of these enlightenment thinkers tend to create their arguments in terms of the simplest form of what man can follow/be, and thus, they lose part of their argument in terms of being able to appreciate how hard it may be to ensure that all people are using reason at a similar level.
    The use of AI makes a perfect example of this; the first differentiation that came to my mind was how machines are more logical and evidence based, whereas regardless of how much evidence humans have, we are still an empathetic species. Overall, great comparison and point!

  2. Nabi, I think you made a great point with revealing the limitations of Kant’s argument. It made me realize that forms of reasoning may differ between humans, as well, since such a drastic form of reasoning can be seen between man and AI. It would be reasonable to believe that every person reasons things differently from one another. This may cause issues with universally applying Kant’s argument on moral law because everyone may not agree with what is considered moral or not. It would certainly be a pain and practically impossible to consider every gray area.

Leave a Reply

Your email address will not be published. Required fields are marked *