simsa03That's why truck drivers navigating their trailers through narrow alleys and plumbers kneeling under the sink will stay employed, and mathematicians and accounts will not.
But the underlying difference is not the one between "know that" and "know how", that old-fashioned distinction between propositional vs practical knowledge. It is, rather, one in the "object known", that is, whether the object is one displaying structure or presenting itself as unstructured.
Nor is the difference between opining alone and knowing together, that is, the difference between what happens when one is, so to speak, "epistemologically alone" and when one is "epistemologically in a team".
One person alone cannot know, as self-correction will only re-iterate the procedure it applied when reaching its conclusion in the first place. Bias, restriction in scope, disproportionality in emphasis... these factors enter, get visible and corrected only when several people engage. (Or when the same person, alone, performs these acts over a large stretch of time.) Thus, one person, alone, cannot know, and a community of people may still err.
With AI entering the stage, these differences become less important.
For one, It acts in a way that is compatible with both, propositional and practical content. It does so in the old-fashioned way humans did, via argument, logic, proof, as in ways that blurs the line between styles of reasoning, between "know that" and "know how". The machine may counsel, give therapeutic advice, differentiate layers of meaning, provide comfort and encouragement. In that way, AI can deal with objects of knowledge that have been generally been the purview of practical knowledge.
(Therapeutic machines are possible because oftentimes human psychological problems are very similar and can be treated via similar stimulus, response, and feedback. People are that simple and their problems can be standardized which is why DSM-5-TR and other catalogues are even possible.)
As AI becomes ubiquitous and pervasive, epistemological change becomes more apparent. It includes the amalgamation of formerly distinct "objects of knowledge" (propositional vs practical), of ways of knowing them ("know that" vs "know how"), of the passing on of mastering them.
With that comes what may be called the fourth narcissistic insult humans had to suffer:
• cosmological (the earth is not the center of the universe) • biological (man stems from the animal realm) • psychological (the majority of personality rests on subconscious factors beyond conscious control) • epistemological: Man is no longer needed for there to be knowledge, art, craft (or societies, for that matter).
In fact, chances are that there no longer needs to be some sort of self for consciousness, intelligence, even personhood to exist. What a bummer.
But not only does hat change the self-image humans paint of themselves. It also changes the way they are still a peculiar (not necessarily necessary) part of knowledge, of craft, art itself.
When the distinction between objects of knowledge as of ways of knowing them are dissolved, and when we take into account the way information and knowledge appear in the context of AI, a different image (perhaps even: concept) of knowledge may lend itself.
Knowledge no longer stays "justified true belief" (with the usual caveats) or the professional application of a skill, both with their respective criteria for what is counted as error and how to detect them.
Knowledge can now become the specific ability to orientate oneself in a landscape, in contexts of information of principally endless, but at a given moment stable, degree of complexity.
Learning then becomes the process of acquiring the ability (or skill) to orientate oneself in such a context; forgetting becomes the loss of the capability one was formerly able to perform.
Which is: To know becomes similar or analogous to strolling in a landscape, the landscape resulting of and from information, history, proofs, etc. To know then means to be able to wander around in a landscape, to have a feel for the surrounding, and when lost to quickly find one's way out.
The distinction between unstructured and structured processes mentioned at the beginning is something that at the bottom points to a richer concept of knowledge, a more humane form of knowledge, one that is not necessarily privy to humans but at which they excel very well.
AI takes away a lot from humans. Which is not necessarily a bad thing. It takes away a lot of self-deception too (as we may now file it). But to call it self-deception already means that humans are far more than data crunching machines. And that our self-image doesn't rely on this.
All GNU social JP content and data are available under the