What are the implications of the statement "If you can't tell, does it matter?" in relation to AI?

Question

"If you can't tell, does it matter?" was one of the first lines of dialogue of the Westworld television series, presented as a throwaway in the first episode of the first season, in response to the question "Are you real?"

In the sixth episode of the third season, the line becomes a central realization of one of the main characters, The Man in Black.

This is, in fact, a central premise of the show—what is reality, what is identity, what does it mean to be alive?—and has a basis in the philosophy of Philip K. Dick.

• What are the implications of this statement in relation to AI? In relation to experience? In relation to the self?

Answer 1

The question in this video is

Are you real?

What does this question really mean? Is the guy asking whether the apparent female (I don't know if she is a cyborg or not because I did not yet watch the TV series) is a human? So, is "real" a synonym for "human"? If that's the case, then the first implication (in the form of a question) of the statement

If you can't tell, does it matter?

in relation to AI is

1. Can we create an AGI that is sufficiently similar to a human that we can't tell whether it's a human or not (by just normally interacting with it)?

Of course, it's not clear what we mean by "normally interacting". As far as I remember, this issue is also raised in the film A.I. Artificial Intelligence, where the AI (the kid interpreted by Haley Joel Osment) looks sufficiently real to the other kid, so he behaves as if he was a human kid, but then the human kid understands that he's a machine, and starts to behave differently (I hope I'm remembering the film correctly).

So, the second question that we could ask is

2. Once we understand that it's not a human (for example, because it's made of other substances), would we humans start to behave differently and start treating the AGI differently?

As opposed to the first question, which is still an open problem, this second question can probably be answered by looking at our relationships with other humans (or entities, such as other animals). Often, we have an idea of a person. Once we discover something new about that person, which maybe we dislike, we may start to treat that person differently. I think this would very likely also happen in our eventual relationship with a sufficiently advanced AGI too, as depicted in the mentioned film.

Now, let me try to address the other question

What are the implications of this statement in relation to experience, in relation to the self?

I think that you're asking whether a sufficiently advanced AGI could be considered conscious or not. Of course, this is a very hard question to answer, because we still don't have a clear definition of consciousness or we don't yet agree on a standard definition, so I don't really have a definitive answer to this question. However, if consciousness is just a byproduct of perception and the ability to understand the world and its (physical) rules, then an AGI could be conscious (in a similar way that humans are also conscious). However, consciousness may not actually be necessary to correctly act in the world. In any case, the AI probably needs to know that it has a body and that it needs to protect it for its survival, if that's its main goal.

Answer 2

I haven't watched Westworld from the beginning to the end, but I've read the synopsis of it.

The implications of the statement above might be related to the experiences of what the androids (or cyborgs) might think of themselves. Are their identity and experiences that they had gone through "real" or not? It seems that, in the series, this question is actually central, and the series seems to be about an identity crisis, the search for a true identity.

In my view, without really any fail-safe programming, once the androids understand that all of their experiences are false (or unreal) and that everything that they do is monitored and tested by some individuals for science purposes, the AI would go rogue. A conscious being with no real background experiences being put into an environment that 'should' have been their 'life' all along, but, after realizing the real truth, everything does matter, and the conscious being would go rogue to find their true self in order to recover their origin and their true identity.

Answer 3

It's useful to understand HBO's Westworld as an extension of Phillip Dick's Do Androids Dream of Electric Sheep.

Most of Dick's novels involve the nature of reality in relation to perception, and how that informs identity.

A major feature of Dick's book, and BladeRunner, is a form of Turing Test (Voight-Kampff) to which humans are subjected to determine if they are replicants.

At a certain point, Deckard, the hero, begins to question whether he is human or an android. (This is never fully made clear in the book, and Deckard's wife's alienation from him may indicate his non-human status. The film adaptation similarly raises this question over potentially implanted memories, which would mark Deckard as a replicant, even though the director reversed course and later stated otherwise.)

Westworld continues this idea where there are characters who turn out to be definitively androids, such as the Man in Black, who, presumably, has had artificial memories created, and believes he is "real". BladeRunner 2049 also involves this theme, which could be said to be the "unreliability of memory" in relation to experience and identity. Even in mundane circumstances, two humans can remember the same event differently!

• The point of the Electric Sheep hypothesis is ambiguity—we can only validate our own qualia, and even that is not entirely reliable due to the nature of perception and subjectivity.

The novel ends with Deckard finding a frog in the ashes, and initially thinking it is real. It turns out to be robotic, but Deckard ultimately decides it doesn't really matter.

• This is important because empathy is the main theme, and altruistic behavior in nature is supported by evolutionary game theory.

The central plot device is that replicants don't have empathy, a design flaw that becomes a "feature not a bug" in that it keeps replicants from banding together to overthrow their oppressors. But the new generation of Nexus androids are intelligent enough to develop empathy naturally.

Dick was a Christian philosopher who worked mainly in popular narrative and believed empathy is a natural function of intelligence sufficiently advanced.

• If the suffering witnessed by an entity appears real, but we cannot validate that entity's qualia, is it a moral imperative to make a Pascal's Wager?

i.e. err on the side of caution and compassion, just in case the entity is conscious.

• If altruistic behavior is expressed by an algorithm, is that altruism invalid?