Tangent - if you haven't read Holt's "how children fail", it's a must-read (for teachers and parents alike).
There's (I believe) plenty of value in exposing children to other children for socialization, but I'm not sure this needs to be at the same time as they're learning, rather than via play and activities designed to promote interaction, rather than learning per-se.
Such a system could effectively flip teaching where teachers facilitate and help with questions while the software ensures the "best in class" content is provided and assessed.
I still think about building that from time to time, but life and work don't leave me with the time to start building that. 10 years ago, I got a little ways into the engine for it for an algebra course and that was it.
(Apologies that it is only in Dutch).
The adaptive software pushes each child optimally for their individual learning rate. Their teachers can monitor their progress at a very granular level and give individual attention in a targeted way. The overall progress of the class can be compared with all other classes across the country.
(Of course, now I'll get a bunch of teachers commenting that they believe in nurture over nature, but c'est la vie.)
So how do you teach to a class like this? The current trendy solution for this is supposed to be 'differentiation'. [1] There's a lot of fancy research and terminology associated with it, but it's a very simple idea - approach topics in a way appropriate to the aptitude of the group whom is learning. It's an extreme burden on the teacher and there's scant evidence of it providing any benefit.
However, tracking [2] is another solution. Tracking is essentially dividing a single group of children based on aptitudes. And unlike differentiated instruction there is substantial evidence of this in both directions. By both directions I mean that above average aptitude students tend to overperform even more when also surrounded by other above average students. But there is also substantial evidence showing that these students start to underperform once placed back into educational environments with lower performing students. The main argument against tracking is are primarily related to equality of result instead of equality of opportunity. E.g. - aiming for everybody performing the same as a goal, rather than everybody performing to the max of their individual potential. That is, in my opinion, very literally a 'Brave New World' style dystopia - hamstring overperformers to try to bring everybody down to the same scale of performance.
And so tracking gets back to the question at hand. When you jam everybody together, 'gifted for everybody', it's not the case that you retain the same level of achievement with even more people able to express their aptitude. Instead you end up a situation where everybody starts getting pulled down by the lower performing individuals. In effect instead of the desired result of "magically seeing more gifted children" you may very well end up seeing fewer.
[1] - https://en.wikipedia.org/wiki/Differentiated_instruction
[2] - https://en.wikipedia.org/wiki/Tracking_%28education%29