(2023-02-18) Rao Text Is All You Need

Venkatesh Rao:Text is All You Need. (This essay is part of the Mediocre Computing series.)... I guess we have our first significant, year-defining news of 2023...the initial reactions of the Bing “Sydney” AI (LLM) chatbot.

In brief, it appears that Sydney has somewhat different machinery under the hood than ChatGPT, and the transcripts suggests a personality that is about the same in terms of coherence, but a wild leap beyond in terms of charisma and colorfulness

I have begun to form some conclusions, not about the technologies, but about people. In particular, about the curious fact that we seem to be displaying an extreme reaction not to computers wiping the floor with us in some exceptional performance domain like Go or chess, but at being completely mediocre and flawed.

We are alarmed because computers are finally acting, not superhuman or superintelligent, but ordinary.

What’s new is that they’re starting to be good at being ineffectual neurotic sadsacks like us in domains where “better” is not even wrong as a way to assess the nature of a performance.

there are billions whose identity revolves around, for instance, holding some banal views about television shows, sophomoric and shallow opinions about politics and philosophy, the ability to write pedestrian essays, do slow, error-prone arithmetic, write buggy code, and perhaps most importantly, agonize endlessly about relationships with each other, creating our heavens and hells of mutualism.

In case you weren’t keeping track, here’s the current Copernican Moments list:
The Earth goes around the Sun,
Natural selection rather than God created life,
Time and space are relative,
Everything is Heisenberg-uncertain
“Life” is just DNA’s way of making more DNA,
Computers wipe the floor with us anywhere we can keep score

There’s not a whole lot left at this point is there? I’m mildly surprised we End-of-History humans even have any anthropocentric conceits left to strip away. But apparently we do. Let’s take a look at this latest Fallen Conceit: Personhood.

personhood I mean what it takes in an entity to get another person treat it unironically as a human, and feel treated as a human in turn

This is obviously a circular definition, but that’s not a problem so long as we have at least one reference entity that we all agree has personhood.

In Martin Buber’s terminology (ht Dorian Taylor for suggesting this way of looking at it), X is a person if another person relates to it in an I-you way rather than an I-it way.

It’s not the fact that chatbots can convincingly present as persons that constitutes the Copernican moment. It’s that they can do so using nothing more than statistically digested text scraped from the online detritus of our social lives.

The simplicity and minimalism of what it takes has radically devalued personhood.

This has been most surprising insight for me: apparently text is all you need

So if you ask what conceit this Copernican moment strips away, it is the conceit that personhood — seeing and being seen in I-you ways — is some sort of ineffable special essence of complex forms of life (salamander and up). It’s not. It’s digested text

let’s back up and try to understand what’s going on here, and why it is that text is all you need to produce this kind of personhood.

Yes, there is a good deal of projection, of a kind going back to very primitive chatbots like Weizenbaum’s ELIZA.

Arguably, projection is how we see personhood in other humans as well.

We can find useful insight in an unexpected place: acting. The closest we normally come to unironically seeing personhood in a non-person is when we forget that a character played by a skilled actor is in fact an invention, forget the actor, and relate only to the performed character.

There are people I’ve known all my life “in person” I feel I barely know at all, and people I’ve never met outside of social media text streams whom I feel I know intimately. And from experience, I know that subsequent meetings “in person” generally validate the online text-based mental model of the person.

Text is all you need to see the personhood in real humans too.

You don’t need embodiment, meatbag bodies, rich sensory memories. This is actually a surprisingly revealing fact. It means we can plausibly exist, at least as social creatures, products of I-you seeings, purely on our language-based maps of reality. Language is a rich process, but I for one didn’t suspect it was that rich.

Still, even though text is all you need to personhood, the discussion doesn’t end there. Because personhood is not all there is to, for want of a better word, being. Seeing, being seen, and existing at the nexus of a bunch of I-you relationships, is not all there is to being.

There’s a lot of nebulous territory around the word “being”

For this essay, I want to limit myself to a narrower question. What is the gap between being and personhood? Just how much of being is constituted by the ability to see and be seen, and being part of I-you relationships?

The ability to doubt, unlike the ability to think (which I do think is roughly equivalent to the ability to see and be seen in I-you ways), is not reducible to text. In particular, text is all it takes to think and produce or consume unironically believable personhood, but doubt requires an awareness of the potential for misregistration between linguistic maps and the phenomenological territory of life. If text is all you have, you can be a person, but you cannot be a person in doubt.

Doubt is eerily missing in the chat transcripts I’ve seen

Text is all you need to be a person, but context is additionally necessary to be a sane person and a full being. And doubt is an essential piece of the puzzle there.

But the most significant part of the gap is probably experiential dark matter: we know we know vastly more than we can say.

This is why I so strongly argue for embodiment as a necessary feature of the fullest kind of AI.

A more interesting question concerns the state of being-together. Can you and I together, between us, in an intersubjective mode, know more than the sum of what we can say to each other in text or text-equivalent?

Here, I must say, I am inclined to the somewhat existentialist view that there’s no there there.

I think sexbots you can marry and live happily ever after with are not that far away. Being-together is likely reducible to an entanglement of mutual-seeing personhoods, and therefore text might be all you need to produce it.

The most surprising thing for me has been the fact that so many people are so powerfully affected by the Copernican moment and the dismantling of the human specialness of personhood.

Either piles of mechanically digested text are spiritually special, or you are not.

The fact that we routinely use an apparently impoverished vocabulary of emoji instead of sending authentic facial expression selfies to each other reveals just how textualized personhood is.

In a quick inventory, based on my either/or pair of choices above (either piles of digested text are sacred, or your personhood is no more sacred than your hairline or weight), I have noticed that almost everybody is choosing the first alternative: treating piles of digest text as having personhood-sacredness, even if it is very stressful to do so.

Strong mutualists (often Illichian conviviality socialists as I’ve come to think of them), whose entire identity is about I-you seeing-and-being-seen games and rituals are desperately scrambling for more-than-text aspects of personhood to make sacred (this is the moving-goalposts crowd of personhood defense).

somehow, I try and fail utterly to participate in any of these unironic strong reactions.

I suspect I am not just choosing the second alternative — there is nothing special or sacred about personhood — I am untroubled and unconflicted by my choice.

Like that bit in an episode of The Scary Door within an old Futurama episode, we will start outsourcing more and more of our personhood duties to AIs, including experiencing the ultimate irony of that.

But there will also be a more generative and interesting aspect. Once we lose our annoying attachment to sacred personhood, we can also lose our attachment to specific personhoods we happen to have grown into, and make personhood a medium of artistic expression that we can change as easily as clothes or hairstyles. If text is all you need to produce personhood, why should we be limited to just one per lifetime?

I can imagine future humans going off on “personhood rewrite retreats” where they spend time immersed with a bunch of AIs that help them bootstrap into fresh new ways of seeing and being seen, literally rewriting themselves into new persons, if not new beings.

I think what is left is the irreducible individual subjective, anchored in dubito ergo sum. I doubt therefore I am. And perhaps that too will come under assault, and crumble, in the face of further advances in our lifetime

and we can finally ascend, en masse, to enlightened nothingness as a matter of routine existence, like the late-stage sublimating species of Iain M. Banks’ Culture series novels.