I really think a lot of the pro-AI people making these bad faith arguments don't know anything about algorithms, let alone the (ahistorical) association of a relatively simple classifiers with a seq2seq transformer