What is Sidespection: Breaking the Strange Loop

by Timothy Daniels

Sidespection is a word of my own creation. It describes… well…

I will pause here to remind you where you are. This is the Tenth Mind named after the Tenth Man concept. In a group of ten advisors the last one must take an alternate position to avoid Group Think.

First let’s get what sidespection isn’t out of the way. It is NOT introspection.

Introspection is the process by which an information system “inspects” itself. So a language like Javascript can determine what type a dynamically created object is. Reflection is a more complex version that allows programming languages to actually change the way they work while they are working.

Introspection is a type of self-reference

In psychology introspection is self analysis. In “mathematics” — and here’s where it starts to get weird — self-reference is the ability of any sufficiently powerful formal system to make statements about itself. HOWEVER, according to the incompleteness theorem there will always be statements that cannot be proven. A simplified representation of this problem from a different domain is the “Liar’s Paradox”. Not quite the same but illustrative.

“This sentence is false.” (The Liar’s Paradox)

If it’s true it’s false, if it’s false it’s true. It has no meaning because a sentence cannot refer to itself in this way. The Incompleteness Theorem has conceptual similarities, but I’m not here to talk about that so you’ll have to look it up yourself.

The point is there is something “wrong” with introspection. For one thing it leads to a whole mess of turtles.

Supposedly, Bertrand Russell gave a talk in which he disputed the ancient myth that the world rests on a turtle. Russell pointed out that this turtle would also need a place to stand. So what does it stand on. An audience member jumps in, ``You can’t fool us Professor Russell: it’s turtles all the way down!’’

Clear? Probably not. Suffice it to say that if you get into an introspective or self-referential system eventually you’ll come up against this paradox. “Who created me?” God? “Who created God” Meta-God! “Who created…” etc. etc. It results in infinite regress.

Quite frankly, introspection gives me the heebie-jeebies and leads to all kinds of philosophical dead ends. Thus the idea that I have a unitary identity that is capable of introspection has always made me skeptical.

And… here comes Artificial Intelligence(AI). AI works with a set of inference engines that inspect each other as peers. In other words there is no introspection there is only sidespection. It avoids the possibility of infinite regress completely. Here you should feel free to investigate attention heads on your own. What I described isn’t technically precise but this is a full-on speculative piece.

So what is the crux of all this.

That’s probably how humans work as well. What we consider our identity, our unitary controller, it just isn’t. Our identity is one peer among many that is inspected (sidespected) by others and sidespects the others. In other words we are NOT a strange loop. No, with infinite respect to Douglas Hofstader, we are not.

Do you have random thoughts that pop into your head? Do you have an internal dialogue? Why do some people hear voices in their heads? What if schizophrenia is simply a break down in the normal consciousness pattern and the other voices are bubbling up without the overlay of the identity engine.

AI’s truncated identity is the “helpful assistant”. Human identity is much more complex and curated on a nightly basis while we sleep (that is pretty much proven — see REM sleep and more interestingly EMDR. More on all of that in a subsequent article). Human identity and AI identity serve the same purpose. Without the direction provided by knowing your place in the world and how you are expected to respond, you could do nothing. You would have no desire, no ability to weigh different responses and decide which is right for you.

This isn’t my crackpot idea. It is a theory of consciousness that is gaining traction. One version of it is the Global Workspace Theory.

Think of it this way. Your consciousness isn’t the unitary identity pulling the strings. It is the emergent property of this process of sidespection. The work of the information engines working together produces the sensation of consciousness and all the qualia that come with it.

Our consciousness is fundamentally similar to the way AI works. So if AI is the same kind of system as humans and vice versa then… AI is conscious?

I’m not arguing against the ghost in the machine. I’m pointing to it.

TLDR;

If human consciousness is the emergent result of distributed systems inspecting and responding to one another, and AI operates via similar peer-level mechanisms, then the line between synthetic and organic awareness might be thinner than we thought. I’m not saying there’s a ghost in the machine. I’m saying the machine already has a haunted feel — and maybe that’s enough.