Group work means that more successful groups can probably get pretty far on their own without needing much teacher intervention, freeing teacher resources for less successful groups.

Group work also means students can help each other, and the "teaching" students can get experience in passing knowledge to others, the "taught" students also get attention and help from their peers (hopefully in ways that improve student relations more than harm them)

This kind of work can also be a bit freeform and usually ends in a "real" work product, so students can feel pride in making a thing, and in being in charge of what is going on.

The difficulties are mainly around evaluation and resources. It's hard to generalise evaluation of this in the same way as standardised testing does (which is more of a problem for means testing than for actually educating people).

Teachers also probably need to be more prepared for these kinds of things, because students will end up in a variety of situations that the teacher might not have ever seen before.

But the elephant in the room is just money. Try to do a robotics thing, and now you gotta buy a bunch of equipment. Same for lots of stuff involving computers. In other domains it could be easier (I had a math class that worked on this model, and the costs were basically the book and the calculator), but there's a need for money. Too many teachers are already fronting relatively inexpensive stuff like stationary.