I think you are quick to judge the talks by their title. They are all technical from what I can tell, and many cover topics that are of interest to PhD research

I think you are quick to judge the talks by their title. They are all technical from what I can tell, and many cover topics that are of interest to PhD researchers. For example, the distributed large model paper covers several papers' worth of research done at Google and how to utilize their implementation in JAX. ML debugging is just one topic among many. I wouldn't dismiss the entire lineup because Google didn't have any big announcements in that space for that year.