[rjsw]

People that I'm currently working with are using AI to try to extract data from the text of published papers, getting access to raw data sets doesn't seem to be a priority.

[anamax]

The data is supposed to be available from the authors in almost all cases, but in many (most?) cases the authors won't provide it.

[autoexec]

It should be easily accessed without even having to ask (links should be automatically provided and maintained by whatever entity published the paper).