2 Comments

I liked your framing of AI systems being used in contexts where the meaning-making has already been done by humans.

For data management would that mean these are equivalent - "...but only where the data to be managed is both explicatable and already explicated." to "where the human meaning-making of the data has already been done"? For eg. in your example, the marginalia has been typed up separately.

Expand full comment
author

absolutely.

and humans need to do a lot of meaning-making work when dealing with data. for e.g., deciding that measures are equivalent ("these 4 countries measure 'literacy' in similar enough ways that we can treat them as using the same literacy measure.") or deciding how many categories to slice the world up into and what those categories should be ("the colour variations in the leaves of the tobacco plants are irrelevant because all the plants are from the same species" vs "the colour of the ripe berries of the coffee plant affects the flavour of the coffee, so we will distinguish between the different berry colours even though they are all from the same species.")

Expand full comment