"A transformer," writes Hollis Robbins, "performs a four-step operation: it takes an input, selects which features of the input to attend to, weights those features based on patterns learned from training data, and generates the most probable output." Aggregate, remix, repurpose, feed forward. Doesn't need to be 'most probable'; it's usually 'most relevant' or 'most salient'. But I digress. "Since 2000," continues Robbins, "American universities built an enormous infrastructure around a mode of instruction that performs the same kind of operation: converting particulars into categories and generating outputs from learned patterns." That's not exactly the same, but again I digress. The differences aren't really significant. Then finally: The university's scaling operation succeeded. The millions of graduates carried the four-step operation into every American institution... (but) The same formal property that allowed one compliance strategy to work across every discipline allows one machine to perform the operation across every institutional context." But here's the problem: "The university faces a market problem. Does anyone need to pay tuition to learn an operation that a machine performs competently?... I can't see any other future but collapse."
Today: Total: [Direct link] [Share]
