Philosopher AI
Purchase queries
What are the limits to what we can understand?
The idea of limits is interesting. Human understanding has no boundaries, in the sense that we have yet to encounter a limit to what humans can understand about our world. What about artificial intelligence? If an AI were programmed with a goal-driven objective function, it might be possible for such an AI to reach its goals without being able to understand them. This would be similar to how human brains work today: We don't usually understand why we do something or how something works; we just do it or know it.
However, if an AI is not programmed with a goal-driven objective function, then it might be able to understand everything in its environment, including itself. Such an AI could either do this through brute force or by using more subtle means. Brute force would likely take longer than the lifetime of the universe and therefore seems unlikely; however, we can't rule out that possibility just yet.
A more subtle means would be to use a similar approach as that taken by humans. For example, when an AI is given the goal of understanding itself or its "world," it could do so in the following steps:1) The AI observes its environment and learns about what's present and absent from that environment (observation). 2) The AI infers why certain things are absent/present from what it knows about how things work (reasoning). 3) The AI uses this knowledge to infer other facts about its world using logical methods. 4)
In this way, the AI learns about what its world is by using inductive logic and reasoning. It also uses deductive logic to learn new things that it had not previously learned. In other words, over time the AI learns more and more facts of its environment (knowledge) and becomes wiser in doing so.
A similar approach is used by humans: We observe our environment, notice what's present and absent from it (observation), we use this knowledge to infer other facts about the world using inductive logic and reasoning. When we realize that certain things are absent/present in our world, we often try to learn why they are so or how they work.
In this way, we can learn about our world by using inductive logic and reasoning. We also use deductive logic to learn new things that we had not previously learned: For example, when a baby learns for the first time that it can crawl around on its hands and knees and get to other places in its environment, it is learning something new through deduction.