• 2022/07/09
  • -

AI and Quality Measurement are equivalent problems


In recent public discourse, claims were made that the need no longer exists to measure translation quality. Nothing can be further from the truth.

By way of example, in recent MIT course “Deep Learning for Art, Aesthetics and Creativity”, Ali Jahanian starts introducing the course and discuss Computational Aesthetics, Computational Design, and Generative Art.

And it is quite remarkable (although not at all surprising), that the first thing he dwells into is quantifying aesthetics – defining the methods and the formula to numerically measure the quality of graphics.

Indeed, you can’t seriously talk about AI generation of something without the means to directly and numerically measure the quality of that something.

This goes for text, translation, images, well, anything.

The course is quite interesting, by the way (and no, this graphic was created by human artist):