Issues of "what is grinding" aside, one obvious solution to "repeating the same content" is to add more content. Now adding content is expensive, and takes time, so how do you work around that? Well, games throughout history have found many ways to get more from their content:
Same enemy, different skin, different bullets. A technique passed down from the very early ages. Why have just one orange "Hill Orc" when you can have red "Fire Orcs" and blue "Ice Orcs" and green "Forest Orcs"?
Cons: Unless you really vary up the strategies for them, most people will see killing red orcs as largely the same to killing green orcs. So...
Grinding is usually considered mindless, so add some mind to your encounters and they won't be grinding. If Fire Orcs behave in a different way from Ice Orcs, then encounters with both will feel fresh. Implementation varies from giving them different attacks to actually giving them different behaviors.
Cons: AI is a lot of work! Plus most MMOs don't have processing to spare on AI. Also...
Vary up the landscape, make encounters that actually use the terrain, then take advantage of that. Orcs in tight hallways should feel different from orcs in wide open fields. Let specialized monsters take advantage of terrain, like Fire Orcs are immune to fire so they can attack from unexpected directions in a firey landscape.
Cons: But all this stuff still requires a lot of content and players will eventually burn through any amount of content you make. WoW does all of these things AND MORE and there's still not enough content to completely eliminate repetition. Which means...
Procedural Content Generation
Will Wright talked heavily about this prior to Spore coming out. Unless you have a 100 person art and design team cranking out quests and assets, its very hard to have a lot of content. The solution is to have a little bit of componentized content and then to mix-and-match the pieces.
Any game with procedural content is a good example for this: Minecraft, Noctis, any roguelike, etc. Look at Nethack: that's a 30 year old game that's still popular. Yeah it gets updates pretty often, but people come back again and again because the randomized, generated content is always fresh.
This is a strategy that plays well with the above too. Nethack has a lot of predefined monsters, but the terrain varies heavily. Different mixes of monsters, in different terrains, require different strategies depending on the player's class, race, and item loadout. The number of permutations means that almost any encounter will be somewhat unique, but still building on everything the player has learned to that point.
Ahh, but the downside is that procedural content is HARD to get right, and can easily shift from awesome to boring. Who cares if you have billions of miles of generated terrain to explore, if all of it is empty and boring? Likewise look at Spore: lots and lots of monsters, including thousands generated by players and available as automatic downloads. Lots of varied and interesting worlds. But only about 3 AI routines so no matter how cool a monster looks, they all play exactly the same. Also a very limited selection of powers and abilities. Every game plays out exactly the same, just with different skins.
Procedural Content is only good when:
1) it is constrained to make sense and be playable ("I spawned in a sealed room full of lava?!")
2) it is actually varied enough to be fresh and the player's don't see the illusion that all of it is just new configurations of the same old crap
3) the rest of the game is built to support and reinforce the value you gain through procedural content generation. Your game has to build on top of the PCG assets, not just rely on PCG to provide gameplay.