Data is: the probabilistic correspondence between systems.
everything in existence, including existence, is a system. And every system is in correspondence with one or more systems, even if only itself and all of existence.
When someone does data science or ai or deep learning or pattern recognition they are looking for higher probabilities of correspondence between systems. The better the data science the more systems are found to correspond with high probability.
For example, gravitational waves of colliding black holes correspond to sensitive instruments in multiple locations on earth that correspond to multiple software programs analyzing the signal of those systems. All this corresponds to human theory of black holes and, perhaps, if we’re lucky something that may actually be like a black hole.
No where in any of that is a single thing called “data”. Nor is it ever perfectly clear which artifact might be a source of noise or probability. That is, when you do data science on blackholes you are all at once observing and controlling for human theory, computer hardware, electro magnetic waves, gravity waves, software etc.
Another example, when someone does art history of paintings one is doing a data science of theory, materials, associated writing, biography, analytic geometry, chemistry etc.
The underlying methodology of observing blackholes or figuring out what a painting is about us functionally equivalent.
But… for different systems of use in human society black hole science might be more efficient than art history or vice versa.
The question before us, as present humans, isn’t whether religion or art or science is the true or best depiction of reality. The question is what mashup of observational and analytic systems improves our individual and collective abilities to find correspondence between the conditions we find ourselves in and those we desire to be in.
The question of ai is not different. As per the above it could not be any different.
So… do we all think we’ve thought through and come up with the best mash up of observational systems to keep ai from continuing to do current human things like strip the earth of national resources while distracting itself with everything else?