@mirabilos
Nowhere in that text do I say it is perfect or resembling the quality of a human doing the same thing. In fact, I explicitly state it gets things *mostly* right "about 95% of the time".

Mostly is not entirely.
95% is not 100%.

Also, these are gut-feel percentages, not statistically sound ones.

Don't quote me out of context, please.
@zwangseinweisung

@mirabilos
Possibly, I'm not hung up on the percentages. A lot also depends on the model being used, of course; some have better results than others.
@zwangseinweisung