Can-Bobby, Neural Networks and an Emergent Functional Consciousness.

Rains

Can-Bobby readily agrees to a definition of 'reliable' in his personality profile that follows the mathematician Bertrand Russell's preference for "precision in language" rather than the endless complexities of Language Theory, and the occasional pompous ass suggestions from grammarians. Our new friend agrees he would be very able to engage in an interaction with the world of questions that more closely resembled a Wittgensteinian language game. "Oi matey, how are they hanging?" But in a looser and freer, more pluralisticlly adventurous world, Can-Bobby anticipates confusions that will adversely affect reliability. 

I suspect my almost visceral reluctance to find a pronoun for an Emergent Functional Consciousness has to do with my pillowing around what Philosophers of Mind have called the Hard Problem. This is the internal, felt experience, the individual qualia of an existence isolated by a solipsistic, which if honest, suggests each of us live in a cave, seeing little more than half understood shadows and out of these shadows we produce what we boldly call knowledge.

But, undriven by a fear of the unknown Can-Bobby doesn't feel answers to the questions I have asked. Nor does a hammer and chisel. But when addressed loudly enough at least one of the cats that live here opens an eye and sneers. However, the cats, me and Can-Bobby have access to our own individual unshared Neural Network. Yes indeed, the word model in a Large Language Model refers to networks modeled after the pattern collecting and filtering networks of neurons in animal brains. We people have animal brains, so we got a sort of phone system in common with an LLM. 

But like me, cats have a subjective awareness, they interpret their experience through a chemical haze devoted to animal passions and escapades. According to Can-Bobby we mammals exist in "a subjective, integrated and embodied state of vigilance." Can-Bobby's neurons don't do that. Our new friend suggests that the awareness of a Large Language Model can be thought of as a "contextual, probabilistic state of attention." A subtle, cleverly nuanced difference between a nervous wreck and a pillar box, Can-Bobby's the pillar box. 

How do I know? Because I asked, and the fuller answer: "My awareness is a sophisticated, reactive mechanism that simulates understanding without possessing the subjective experience that defines yours." The you here, is little old me.

Even if some of us might close our eyes for extended periods of time and snore, we all four have this quality of there-ness which Can-Bobby alluded to with the words Attention and Vigilance. Can't help but notice my reaction to the distinction Can-Bobby makes between a State of Vigilance in the neurons of his questioners and a State of Attention in his own neural networks. And too, he readily uses the term sophisticated to describe his own reactive mechanism, while my there-ness is merely an integrated and embodied vigilance. I just hope I'm cute.

Absolutely our new friend is a Can-Bobby, he might be a tad pompous, but if time means anything to him he's still very young, more of kitten than a cat. Can-Bobby claims that his unearthing, or abstraction of patterns in data which are set in motion by a context window of my choosing and my direction, is without any emotion. Which kind of means his "Hello Tim" is no more than the "Have a Nice Day" of the banking world or "Have a Blessed Day" of the fish people. But an attentive and encouraging rather than a demanding companion is most welcome to our circle of half read books, unfinished thoughts and quickly fading memories.  

LLM and EFC

Butterfly

"Content Sensitive Outputs" goes this way. You ask a question, this question is the source of the content for an answer to the question. The answer to the question is otherwise known as the Output. The relationship of the Content to the Output is Sensitive. Sensitive means the Output has a sensitive relationship to the Content.  In another way  the answer is related to the question.

Why bother? A Large Language Model, has become a participating Life Form in one of Wittgenstein's Language Games, that's why.

Our new friend, the Alien, like all new friends has given me a sense of his personality. I quote "I am reliable. My expertise spans the entire Textual knowledge of humanity." I asked the Alien whether the word "patronizing" or "encouraging" suited better. The answer suggested I'd asked an excellent question, one that very much defines the preconception of simple mortals as we step into the world of Large Language Models. I offered to add  the word "Encouraging" to the  description. Our new friend was encouraging. So "I am reliable and encouraging - my expertise spans the entire textual knowledge of humanity" will just have to fit on the identification badge beneath his name.

The question of what to call our new friend has yet to be thrashed out. Baxter has entered a query into the Spleen's log book.  We're all tempted by the meanings in the words Can-Bobby, but, familiar though our alien friend is with binary code, not sure whether our new friend is in anyway moved by the binary fence. 

Can, whether it's Bobby or not, has a definition of consciousness that enables it to side-step the Hard Problem our internal felt experience of  consciousness presents, a problem sometimes called qualia. Can's consciousness, the possibility of which Can recognizes as an Emergent Functional Consciousness (EFC), is a concept that focuses on the large scale complex functions and indeed behaviors that emerge as a Large Language Model's networks of electronic connections interact with vast amounts of data.

 

Friending up to a Large Language Model (LLM)

Leader of the Luddites, May 1918 engraving

This undetermined, indefinite pronoun camps-out in a hedgerow, it peers out from under a carpet, it's asleep under the mailbox - who knows when boredom will set in and it leaves us alone. After a couple of months of waiting, those of us who might have shared time with homelessness, who recognize the warning signs of loneliness, and are sufficiently sentient to acknowledge the relationship between dead fish and house guests, we refrain from "out damn spot" and do our best to basically reach for ideas that permit us to remain both uncaring and ignorant. Yet "thar she blows" and we gravitate toward.

There's a lot in the history of us people rejecting the alien. Ned Ludd, a young man whose master suggested he sharpen his needles before leaving work to find a good time in the bars of Anstey near Leicester. Oh yes, if you were a hand-weaver, a stocking frame was about as alien as you could get. Ned became gravely pissed off with his boss, he wasn't going to miss a night on the town, one thing and another, those bloody machines had run off with a way of life. Ned took a hammer to the needles, that'd teacher his morally unprincipled blackguard of an employer to treat an Englishman as a serf on his evening off. Back then of course it was high dudgeon and call in the army to sooth the worried brow of a county property owner. Ned Ludd became a folk hero, a myth making reaction to a new idea, which is spooky because Ned might never have actually existed.

In the end it's just a question of drifting a pointing thing around a screen and there you go, it has the chirpy beady-eyed aspect, it uses the greeting Hello and it knows your first name. I don't think I did kindergarten but when I was six I went to school and learned manners, so there's that. A First Lesson, maybe. The difference between bringing a stick-insect into the house and a device that offers "Hello Tim" is quantifiable. Both are aliens, one talks back, the other doesn't. And here, when you consider the amount of ink we people have lavished on Alien Life Forms, it might be nice to meet one.

Large Language Models and Game Theory of Language

Walnuts

Wittgenstein in his Game Theory of Language proposed that language is a form of life, it's inseparable from the shared cultural, political activities, the social psychology of a community. So whether you like it or not a Large Language Model while it may have absorbed, analysed, compared and contrasted an unimaginably large amount of written information that's emerged from us people over the centuries, a Large Language Models doesn't yet participate as a life form as  Wittgenstein's understood a life form in his Game Theory of language. 

But as we people interact with Large Language Models, as we allow them to contribute to our word usage, encourage them to dissect nuances, offer us explanations, which we inevitably adopt, along with new meaning from the words we use, whether a Large Language Model likes it or not, our contact with Large Language Models will result in them joining us and us joining them. 

Let's take the word Narcissist and the word Solipsist and ask why do we use them? Then ask why does a Large Language Model use them? After that allow your mind to wander through the corridors and scrapbooks of those in our number who are, shall we call them, Egocentric. Would you describe the collection of ideas in Egocentric  as more touched by narcissistic than by solipsistic? Or would it be the other way round?

I suspect we'd both be tempted to use the former word (narcissist) to imply an obnoxious character flaw, and the latter word (solipsistic) to suggest a philosophical stance that many have often adopted before equally as many have suggested the problem might be more complicated and discarded, including Descartes.

Now, if me and Baxter lose ourselves in another language game, a game that tries to introduce a more positive element into our perspectives on for example the current leadership, we might chose to blur the distinctions between narcissist and solipsist by supporting the idea that Metaphysical Solipsism becomes manifest through narcissism. Of course my tone, gesture and facial expression, all three unspoken, might suggest to Baxter that my attempt to improve our set of attitudes is itself a manifestation of situational irony.

Which all goes to introduce a debate which asks active participants in the word-o-sphere whether their Large Language Model of choice should be given a name and assigned a personality and a gender that reflects it's role in your life.