4 ms·

I'd lean yes -- what data set can they feed into the training algorithm to replace the stuff that doesn't get into writing?