My world is as close as possible to contemporary real life under the constraint that orcas (and maybe other cetaceans) are significantly smarter than humans.
What is smarts? We define it as information processing that machines cannot in the near future do. This precludes tasks humans need their dexterity and six senses for if said task could be within the range of our AI for a robot with similar "hardware". However, it gives a reasonable metric of what kinds of skills will be most needed (and thus most plot-driving) in an increasingly automated society.
How the intelligence stays hidden (for a while): Humans are very interested in orca and in general cetacean intelligence. Human brains have over twice the neurons as elephant brains, but orcas have twice the human level. But curiosity is not enough.
Firstly, the aquatic behavior is hard to understand for landlubbers. Behavior would have subtle uses, and humans wouldn't have the "bigger picture" such as exact location of prey sonar echos, etc to understand what is going on. It's all too easy to see a cultural ritual that isn't your own and mistakenly think of it as primitive.
Secondly, in my world the language sandbags itself. Researchers manage to find some orca words that correlate to simple concepts, and from that it seems that most of the language must be similar. But it's the parts that don't get deciphered that are more complex and involved than even the Navajo language.
Thirdly, captive orcas get trained more slowly than humans because of this language barrier, not because of less intelligence, which could also mislead people into thinking that they aren't as smart.
Fourthly, humans conflate intelligence with technology. The lack of orca tech is due to the difficulties with the aqueous environment, lack of opposable thumbs (or suction cups or elephant trunks), and much smaller population sizes. Not brainpower.
Given these (human) difficulties, how would humans "discover" and utilize this super-human intelligence?