I too think there is merit in exploring to what degree conciousness can be approximated by or observed in computational systems of any kind. Including neural networks.
But I just can't get over how fake and manipulative the framing of "AI welfare" or concern over suffering feels.