I'll go into a few details to supplement the other answers.
Don't let NPCs be dehumanised
Games aren't made in a vacuum. Through our experience with the medium we have been trained to expect certain things, one of them being that certain NPCs are important while others are nameless background characters. We assume that a named NPC has a role in the story or a quest while someone just called "guard" or "merchant" is simply a hull for their function and can otherwise be treated with impunity (Philipp said as much).
If you want the player to treat NPCs more like people, one possibility is to subvert this expectation - either by giving every NPC a name, regardless of how irrelevant they are or - possibly more powerfully - naming every NPC after their appearance until the player character learns their name, thus suggesting that behind everyone's apparent meaninglessness is a person.
Let actions have consequences not just for the player
Everyone knows that when you're caught committing a crime the guards will be after you, but little consideration is often given to the victim. If you break into someone's house unnoticed and rob them blind, they'll now be poor, they may become homeless and starve along with their families. If you kill someone and nobody catches you, their family will grieve and if that person was their breadwinner, they will fall into poverty again. This can have mechanical consequences (the merchant is bankrupt and there's no trader for you to sell to now), but the surprise of a player returning to the city and finding the house empty and the family begging in the streets may have the greater emotional impact.
Keep in mind that this requires a significant investment in terms of development. I would suggest simplifying the reaction as much as possible by creating a few trigger conditions (death of X, net value below Y) and a few consequences (grieving dialogue, possibly just with the names adjusted, new NPC locations with begging dialogue) and try to apply them as broadly as possible. A player who tries to play around with the system may notice that the depth is limited, but the reaction you strive for is for the player to be shocked, feel sorry and not do it again, so the majority shouldn't have this experience.
One game to maybe have a look at for this is This War of Mine.
Create fitting mechanics for the philosophical underpinnings of conflict resolution methods
That's a mouthful, but anyway: the others have already made the point that every method of conflict resolution needs to be mechanically interesting in order to be worth using. Games that have a deep and visceral combat system encourage the player to use it. I would go a step further: try to design the mechanics of each method to make the player feel a certain way according to your message.
For example: If you want your players to try and understand the NPCs in order to convince them in dialogue, take a look at the social bossfights in Deus Ex: Human Revolution. You have a few character types/traits that influence how characters react to certain approaches. A proud character will react positively to flattery, while a fearful one will dismiss it, but can easily be pressured. Essentially what you're creating here is a system similar to turn-based combat with elemental weaknesses and resistances, but you need to actually listen to the characters in order to find out what they're susceptible to, humanising the characters in the process. Take care to give most characters more than one trait, otherwise it becomes formulaic.
If you want to expand on such a system you can introduce more complexity for important NPCs by giving them hobbies ("Did you paint that waterfall in the dining hall? That's pretty amazing!"), relationships ("I know this is a lot to ask Mr. Bert, but Ernie told me I could trust you with this.") and desires ("I know it's a risky proposition, but if you want to win Princess Buttercup's hand in marriage you'll need to convince her father and I'm told he's a big fan of pirate novels"). This further encourages not just talking to the person in question but to explore their social environment.
Similarly, if you want to communicate that solving problems with violence is barbaric and cruel, design your combat system accordingly. Have combat itself be frantic and stressful through high speed and high lethality. Let weak enemies try to run away and miserably beg for mercy. Have injured opponents bleed to death in whimpering agony or even beg for a finishing blow to not lie dying for hours and days (I think there were a few FPS games about the world wars that tried to depict how horrifying these injuries can be). Then reference the point about consequences above for further fallout.
This can still lead to a kind of power fantasy for the player, but one that is stained and dirty. The player may enjoy it in the moment, but they should feel guilty afterwards as if coming from an eating binge and feeling the nausea or having masturbated to some really effed up porn and wondering where their life went (no kink-shaming intended).
Offering options takes lots of work
An unfortunate reality of game development is that resources are limited. If you want your player to be able to talk their way through every quest, fight their way through every quest, have a few solutions for stealth and crafting as well and maybe magic on top you'll quickly run out of time and money. Not every method will work for every quest (some people just won't listen), but even then you should probably decide on a few main methods (e.g. combat and conversation) and use the others as supplements (e.g. stealth to find information to use in conversation). In the end a game is supposed to induce a certain experience in the player and that experience should determine the core concepts and mechanics and the corners that can't be cut.