Do NPCs have any kind of awareness?

[DoPo]

even if it is fully able to think for its self that's only because it was programed to be able to think for its self.

Sounds like the Chinese Room but the application is internal to the AI as opposed to external for the examiners of the AI.

But the puppet in this scenario is the player character, because puppets are controlled.

Nothing controls the A.I. in these NPCs, it's just the A.I. and yes they're reacting to stimuli in their environment with 1s and 0s, but as I said earlier... don't we kind of? The very emotion we call love is...

I suppose we kind of do. But then again, there are various different ways to react to the environment. A wind vane reacts to the environment:

Spoiler: hey, today I learned how this was called

If you give it the correct input (wind) it would react and give you the output (direction). I doubt anybody would claim it's "alive", "aware" or even "intelligent" in any capacity because of that.

But that's too simple, is it not? It's just using physics principles to determine the output. There is no actual decision making taking place. It would be like claiming a rock knows about gravity, because it falls if dropped.

How does that apply to AI, then? Well, I guess the question is how are they reacting to the environment. If they really are just glorified wind vanes would they and should they truly be considered "aware" then?

There is a class of...well, entities called "agents". They sort of fall under AI[footnote]partly because it's really hard for anybody to agree on WHAT THEY ARE. Like what classifies as an agent and what doesn't. And how intelligent truly are they. It sort of follows that if you don't know what an agent is, it's hard to say is it smart or not. Still, agents are recognised as things it's a bit (read: a lot) blurry where the boundaries are.[/footnote] and they vary in the complexity of their decision making.

And I present to you


This is a robotic agent. There are a lot of takes on the concept of this robotic car also and also software implementations of the concept but I like the physical thing since I believe it's more fun to showcase. So, anyway - the Braitenberg vehicle, as you can see moves by itself, and it swerves and it does stuff. Not a lot of stuff, granted, but it isn't a simple RC car you get off the shelf of any toy store - it doesn't take a person to operate it. However, here is the thing - even though it moves by itself, it's super dumb. And I mean it - the principle behind it, is not in any way intelligent. It is very slightly more advanced than how a wind vane works - it has two motors (which move the wheels) connected to photo sensors - the more light the sensors receive, the more power power they give to the connected motors. So, essentially, you shine a light at it and it moves. If you shine a light from one side, it would move to that side[footnote]some implementations cross the sensors, in which case the vehicle appears to move away from light, rather than toward it[/footnote]. That's it, there is still no decision making involved.

Spoiler: Behold

Here is what is under the hood

And here we are. Reacting to the environment doesn't necessarily mean some measure of intelligence. AI can do that as well. Heck, Searle's Chinese room which I mentioned in the beginning of this very post is another example of how an AI may be not truly intelligent.

However, with all that said, are NPCs aware? Honestly, I've no answer to that. It depends a lot on your interpretation of what that means and also to an extent to how the NPCs actually operate. As you can see, it's a bit muddy and hard to figure out.

 

[EHKOS]

No, of course not, not in any sense at all. This is like asking if a puppet is alive. It's an image of a human that appears to act of its own volition.

NPCs are images on a screen that are there because some lines of code said so and 'react' in ways determined by lines of code.

This is like asking if a choose your own adventure book is alive because it 'reacts'. It's just a fancied up choose your own adventure with computer images instead of imagination providing the visuals.

Exactly this. It's a wireframe model with a texture. You could bend the arms into horrible contortions and cut off the head and it would still walk around like normal. It's a clever imitation of life, nothing more.

 

[Secondhand Revenant]

No, of course not, not in any sense at all. This is like asking if a puppet is alive. It's an image of a human that appears to act of its own volition.

NPCs are images on a screen that are there because some lines of code said so and 'react' in ways determined by lines of code.

This is like asking if a choose your own adventure book is alive because it 'reacts'. It's just a fancied up choose your own adventure with computer images instead of imagination providing the visuals.

But the puppet in this scenario is the player character, because puppets are controlled.

Nothing controls the A.I. in these NPCs, it's just the A.I. and yes they're reacting to stimuli in their environment with 1s and 0s, but as I said earlier... don't we kind of? The very emotion we call love is...

They're both puppets. One being controlled by the player, the other being an image controlled by lines of code.

They don't actually exist as a being. There's simply the illusion of a creature being there. Like I said, this is just a fancy choose your own adventure book. It's like having the words on a page accompanied by an image. Someone else wrote out those reactions

Edit: additional thoughts. It seems overly simplistic to compare our reactions to that of a computer. That's like saying a trebuchet is alive because it reacts. It's just reacting to stimuli in its environment too. All physical objects do.

 

[The Rogue Wolf]

In Skyrim, if you walk around without clothing, NPCs will comment to you about it. But if you strip another NPC bare, none of them will say a thing.

So, no, no awareness. Just a routine for an NPC near the player to check if the "body clothing" slot is empty and then spout a prerecorded line if "yes".

 

[The Almighty Aardvark]

People are on the right track with the anthropomorphising comments, though this is an interesting and complicated subject.

There's many, many different entities in the game that exhibit the same automated processes that you're describing. There are level entities that will behave in certain ways, there's object entities, and there's entities that are far more abstract than that. For instance, in the game I made last semester we had entities that would load shaders for how an object would be lit, register events that occured in the world, and smooth controller input.

There's nothing meaningful to distinguish these objects, they all execute programmed tasks, and they're being created and destroyed constantly. The NPCs just have a wrapper around them that looks like a person, and only get destroyed as a result of a player action. If you'd feel guilty about killing an NPC, you should feel guilty about just booting up the game because you're killing thousands upon thousands of entities every second.

It's also worth noting that the player object itself is not necessarily the thing making the decisions. In our game we had an AIManager that would iterate through a list of players and make choices for them based on the situation. So if you destroyed a player object, the thing making the "decisions" would still exist. Though even that's a pretty simplistic view, since the AIManager class doesn't really exist either, it's just a convenient abstraction of sequential binary instructions. The only way that makes any sense to think of this is that the machine itself is the one thinking, and even that's pretty unlikely.

That being said, the root of what you're asking is a pretty interesting question. There's not really a good way of distinguishing between thinking things and non-thinking things. Both the Chinese Room example by Searle (That Dopo mentioned), and the Mary's Room example by Jackson are interesting thought experiments in regards to this. My personal thoughts is that there's something "more" than purely computation that goes into awareness.

Mary's Room is an example of how there seems to be something missing. The thought experiment involves a woman named Mary who spends her whole life in a black and white room getting no feedback of color. In this room she has a black and white computer that will tell her everything about everything she wants to know. She stays in this room until she has learned everything there is to know about everything, including all there is to know about color. After she's done this, the door opens in the room, and she walks outside and finds an apple. Despite knowing everything there is to know about color, what red looks like to her is something completely new to her.

This suggests that there's something about our perception that's separate from the physical world. The appearance of red doesn't exist in the world, it's something that exists entirely in our heads. Data from programs has no interpretation. It doesn't mean anything until we see the output, be it a picture or a a number.

 

[aegix drakan_v1legacy]

Of course NPCs don't have awareness. That's just silly talk.

That said, it's a fun angle to use in a story, and it leads to some really fun lines of thought if you really think about it.

Maybe I'm just biased. I loved the hell out of Undertale and it played with that. And (tactfully predating Undertale by 6 months) I also made an RPG that dealt with that same idea of NPCs having awareness, and I did not hold back any "feels" potential there.

 

[DementedSheep]

IS this a 2am topic? cause it seem like one.

 

[Creator002]

As someone who has programmed AI: No. No they are not. Even the most intelligent NPC is just a masterfully woven onesie with about a quarter of a skeleton inside.

Same boat here. The AI I've programmed is as "aware" as I make it, and that awareness is simply just "if X, do Y."