Yes and no. You can utilize ML to do 90% of the work in preparing an input, then have a human verify to ensure it's correct. It's similar to how ML models are s

Yes and no. You can utilize ML to do 90% of the work in preparing an input, then have a human verify to ensure it's correct. It's similar to how ML models are starting to consume ML generated artwork, which people think is bad, but in reality those ML generated artworks are the ones humans found good enough to publish online, so they still retain value for training (as being appealing to humans).