Immensely important, and core, question. Love how @emilymbender breaks it down for us. Definitely something I wish people talked more about.

What makes something data?

“This off-loading of the concerns of dataset construction to a different group of researchers and culture of disdain for data and those who get their hands dirty with it partly set the conditions, I believe, for work on language technology to come unmoored from scientific foundations.”

“Just because ‘AI’ is the magic word that triggers massive venture capital investment doesn’t mean the rest of science should behave as if magical thinking and poor data practices are acceptable.” https://dair-community.social/@emilymbender/115635213187268659