Semantics
When the meaning of spoken or written language becomes so important that all the other things that can be done with language become less important, this is a tragedy and kind of describes the world we live in today. Not in the sense that everyone cares a lot about making everything very meaningful but in the sense that meaning is considered important to have. Everything is judged in the context of how meaningful it seems to be. If something is absolutely meaningless, if it is pure gibberish, it is considered to be pure noise.
If gibberish is considered the production of pure non-meaning, and we accept the fact that machines produce better non-meaning than us, what does it point to?
In my understanding of capability, if something is very good at doing 100% of anything, it means something about its core capability. If machines can produce perfect gibberish all the time, meaning might be another kind of gibberish for machines.
Gibberish here means anything that lacks an intended meaning. Machines can have no intention, so everything they produce is always gibberish. So artificial intelligence models represent the intentions of many people (this representation makes them seem intelligent) but they are still producing gibberish. They can never intend anything.
So students parroting what they cannot explain or make sense of are narrating gibberish. Fluency is gibberish when it does not match an intention.
This is a sharp point also because of what it leads to: students will not realise that they are narrating gibberish via an ordinary rehearsal with an LLM because semantically whatever they are parroting is not deficient in any way. An LLM can (via representing others' intentions) give you a reading of how what you seem to intend will be received — but if you are pasting one LLM's output into another LLM's input — the reading will not be correct every time, because LLMs can figure out the semantic layer of the input text but the intent does not live there. Intent cannot be semantically validated.
So this means we need to be so focussed on the language that we use. The language will or will not represent the intents that we hold and that is its limit. Interfaces that allow the semantic meaning held in our text to be validated against our held intent (manifested in some unforgeable way) will help us be believable.
Now that we come to forgery, we need to talk about ciphers. Passwords are human or machine produced gibberish that a machine can hold to implement a person's intent in some way. They are still gibberish, significant but meaningless.
The interfaces for unforgeable intent that I spoke of are mostly physical today or ones that represent a physical presence (like video calls) but the scope is wide open to figure and make more such interfaces possible.
Because we have remote presence possible today in some ways, we can try to see in which ways these remote forms of presence can validate the semantic representations of intent.
But separating unforgeable intent from representations of intent is needed first. The two are often conflated, today.
Intent that doesn't have a semantic representation is of much more interest. I bind semantics to verbal language and do not interpret it in a broad sense to have something to do with all forms of linkages that lead to meaning-making.
Language must be understood as an obstruction to understanding forms of intent which lack a stable representation. And this obstruction cannot be wished away — we are often in a predicament where we do not understand how to say something we feel in semantic terms. And these are the moments when we struggle and reach out for interfaces that allow for other forms of interaction.
Such reaching out sometimes is able to find something useful. But often not.
So we invite you to see language as a kind of prison that is enforced via semantics.
In this prison, we all struggle to connect with each other. In the disconnected form, the current capacities to read cues, hints and gestures are the only ways available to break out and try to connect.
We are likely to keep doing this if we believe these things to be true: the prison is real, the struggle to connect is real, the attempts to connect when we are disconnected are genuine.
My shared experience which you cannot validate with yours is either understood as a sign of my derangement or an anomaly. It is only accepted broadly if I am seen to be giving words to what people-at-large feel but have struggled to express. Between these two is what I understand to be a blurry and confusing process — that does not happen in the same way every time. Is this process about the dispersion/popularisation of ideas that includes their being published and validated along the way? Can readers not rely on their own faculty of analysis and judgement?
Can you?