I have a solid understanding what the numbers are. If I want, I can see numbers in everything. Could an AI have the same ability for any incoming information to tag them by numbers same as I have?
A understanding for this level of abstraction is technically possible. It is not hard to create an AI able to count. Unfortunately, this does not imply that the AI knows what 1 is. It knows: This is 1 piece of cake and this is 1 sheet of paper. But the idea of the number is not grasped yet. Also, animals can count, but they do not understand why it works.
I would say that not even every human knows what 1 is: The number, who is the neutral element of the multiplication. The natural successor of 0.
I can say with certainty, that there is no AI, who reached this level of abstraction.
If we had a clear definition of what is meant by understanding, this question would be easier to answer. The question is obviously thinking beyond the mechanical understanding imbued into an arithmetic unit in a CPU to be able to add, subtract, multiply, divide, and compare an integer, one of which is 1.
What other kinds of understanding can there be?
- Conception of counting numbers and that they begin with 1
- Oneness in the Buddhist sense and that once nirvana is achieved 1 is everything
- Notion that 1 can be a representation of uniqueness
- Distinction of shapes that are visual representations of 1 in various languages
In recognition of one in sensory input the above kinds of understanding correspond to the below AI challenges.
- Cognitive skill to the degree that an elementary school student can begin to understand the difference between counting numbers and integers.
- Comprehension of composition and that a person's consciousness may transcend the confines of a single individual's brain networks
- Ability to understand entity relationships as a database schema designer would
- Numeric recognition in multiple fonts and for multiple locales
The top three seem seem far from what AI has thus far achieved. Only the visual representation is available in implementation form and requires training sets that represent the variety of fonts in multiple language locales required by the use cases.
The question did not say, "Does," but, "Could." Since brains do understand at various educational levels these things in the top three examples of kinds of understanding of one, it is likely that some combination of semantic nets, deep learning, fuzzy logic engines, difference engines, and other AI forks of research will some day achieve these higher abstractions in comprehension and be able to apply the understanding at these other levels.
Interestingly, this would mean that computers may someday be able to do these three things.
- Develop math rather than mechanically manipulate mathematical expressions and search for theorems or calculate math that has already been placed in closed form
- Develop spirituality, have conversion experiences, commune with God, and transcend rationalistic conceptions in thinking
- Develop software so that programming jobs would be obsolete