There will always be stupid kids in a world where intelligence levels vary.
In the 80s and 90s some attempts were made to try to escape this reality by changing the question. Multiple-factor intelligence models were proposed and widely accepted by pop-psychology, but most academics acknowledge now that there is some general intelligence level, a "g-factor".
Anecdotal - I always remember my aunt who was a literature teacher. She always stuck to those ideas and told me that it's obvious that there are different types of talents, because there are people that are good at maths, but do a lousy job in her classes. What she did acknowledge once though, was that people that were good at maths were also the ones that have been writing the best essays from all of her students. There was something that underpined their skills in both areas, but since they knew that they can achive something real in maths they probably did not try so hard to "pick the right key" in interpretive literature tasks. On the other hand for those students that were a bit slower literature was the only field that thay could perform well. Good literature students did not have any particular field in maths that they could perform well in. They stuck to literature, because they could memorize all the important answers and boast that they poses some innate talent.
Anecdotal 2 - Sure, there are some kids that could perform in one area better than in the other, but in my experience this variance is overshadowed by the difference between kids that can do ALL things well and kids that can do NO things well.
EDIT: A nice example of the multi-factor intelligence "scam" are the books from D. Goleman in "Emotional Intelligence" series. They were mostly targeted for middle management people and were very in line with the whole trend. The main motive is that emotional intelligence is a much bigger factor than general intelligence in the context of career progression and overall success in life. What was not made obvious is that the comparisons were done in environments of people that were already successful. If you take a group of people that have a high iq level, then it's very possible that the most successful will be the ones that can work with people a little better. Unfortunately it's not sufficient to be a peoples person to achieve high efficiency in such environments. The "hidden" conclusion was: GI + EI > GI; EI < GI; GI + GI < GI + EI (at least to some degree). There were also some unresolved questions - can you really possess EI without a sufficient level of GI? Isn't EI just a form of GI? In summary there was a lot more of nitpicking of definitions than actual science. It all felt like p-value hacking on language level.