I don't get the hype around "small AI".
If it is based on specific datasets, local, even deterministic - that sounds like a traditional algorithm to me.
Calling all tech "AI" is making it harder to talk about the harms of the large stochastic models.
Resisting "AI" does not mean to resist all tech. It means asking what kind of digital infrastructures we want - and rejecting vast extractivist data centres, stolen data and automated inequality while evading accountability as part of the answer.
