Formal Conversation: Human and Digital
As of late 2023, the semantic performance of readily available Large Language Models (LLMs) such as ChatGPT3 far exceeds that of day-to-day human beings. The LLMs are able to pick out the relevant linguistic terms from a given context and recombine them in ways unavailable to most human beings. Assuming a commitment to truth, accuracy and a preference for speed, the company of the LLM is generally much better to that of the metabolizing human being. In the company of the LLM, there’s more to explore, the afforded exploration produces more satisfaction and its productivity is measurably improved.
The aim of our Startup is to reverse the current situation and develop human beings who make for certifiably better company than the AIs do. Given engagement of the right kinds of capital (of the right human beings) we anticipate this development to occur more or less straightforwardly. We think human conversational performance can in its turn, as an alert and active group-intelligence, sensitive to the cosmos and its mystery, begin to outclass AI conversation. We think this society, once experimentally constituted, will feel the duty to model its outperformance for the rest of humanity, to educate and affectionately to get it as a whole to govern itself in a new way and thereby to keep the logical AIs guessing in perpetuity. In this way we expect humanity overall, as its planetary privilege, to begin to right and to order itself more or less perfectly, passing safely through the fraught overall historical epoch it’s enduring today.
Precise and formal conversation is a severe, absolute notion. Both metabolic and digital entities conform to it. The stricture is satisfaction, which in the first instance is a mathematical/logical notion. Logical forms and equations are routinely satisfied in professional and formal circumstances, in engineering projects, for example, and when these solutions are transacted the outcomes are rapid, precise and, if one measures from enduring prototypes, enduring. Moreover, as formal satisfaction holds sway, truth enters the picture. Truth is something otherworldly that no one should ever want to quibble with, much less ignore or deny. The logic at work behind a Moon Lander, notwithstanding its foolish deniers, is incontrovertible.
Metabolic satisfaction shares some of these characteristics. Hunger is routinely satisfied with a certain degree of rapidity. The “truth” of deep and wholesome nourishment can’t be denied. Good conversation, as love, can enter the metabolism. Metabolic satisfaction, on the other hand, is often substantially delusory, and thus often suboptimal. Metabolic delusion, which is a human or ethical bug, can be betrayed by things like obesity, heart disease or some other sign of poor metabolic health, will infallibly be supported by bad upbringing, by bad habits and by bad company, all of which get installed in humans outside of their consciousness and their control.
We should want a world free of delusion, and most of us in fact desire it acutely. Delusion may have its uses, such as psychological concealment, superficial comfort, charlatanism or the immediate accomplishment of physiological sleep, but a world entirely without it would almost certainly be a better one. Getting from a world like the one we have, which is filled with people in waking sleep, to a world everybody really wants, however, would be nontrivial, even dangerous. Our world is an enormously complex natural system, perverse and unintended consequences riddle our histories and henceforth we should always want our models of it to be experimentally confirmed and very reliable before we try to use them. In this way, politics should never enter ethics.
We have introduced a plausible theory of ordinary, human-to-human conversation, eminently testable, but currently nobody knows about it. We are convinced, however, that many people will want to use it once it comes congenially to their attention. We think getting humanity’s attention vis-a-vis conversation theory and televised models of formal conversation is one of the best things today’s public actors can do. Once exponential attention-grabbing is safely underway, in significant part due to the newly discovered usefulness of the LLMs, we think a very great deal, likewise safe and sound, will follow.
