I'd be very surprised if they're genuinely out of research ideas to test in space. If that is actually true then humanity has a problem.
>research that couldn't be done with automation
I'd think there is room for both. Automation makes sense, but don't think the versatility of meatbags is entirely there yet.
Meatbags are versatile but really, really expensive. They require a really vast support system, and it has to be highly redundant because the cost of a loss is so high.
You can send up a lot of less versatile bots for the price of one meatbag.
> if they're genuinely out of research ideas to test in space
A bigger problem is lack of expertise. Astronauts are not specialist in whatever is the topic of the current experiment. You need probably like 5 years of training (assume the second half of the undergraduate degree, and perhaps the first half of the PhD). So experiments must be fully automated except for a button to turn they on and off.