Open Letter To My Students 2025: Stay Statistically Improbable

Dear students,

I could have easily written a prompt to generate this letter with ChatGPT but chose not to. I didn’t want to deprive myself from learning, making mistakes, accruing semantic capital and most importantly the pleasure of linguistic action and the various blisses of language.

What is semantic capital, you would ask? Semantic capital, as defined by Luciano Floridi is “any content that can enhance someone’s power to give meaning to and make sense of (semanticize) something”. And what is linguistic action, you might also ask? Well, in very simple terms, it is using language to create.  To prompt (pun not intended at all!) action. It is something people before us knew very well and have been exploring for years looking to understand. Words are actions. We act by using words. We use language to shape worlds. And to an extent to enjoy its blisses.

Speaking of blisses, have you heard about Roland Barthes?

Here’s a quote by him, as an interlude in this text to ponder on:

“The text you write must prove to me that it desires me.  This proof exists:  it is writing.  Writing is:  the science of the various blisses of language, its Kama Sutra (this science has but one treatise: writing itself)“

Let that sink: “Writing is the science of the various blisses of language”.

It is that bliss that makes a text alive albeit many times messy and waiting for the person reading it to add their part onto. It is that bliss that I see lost in the flat world of me typing something in a dialog box and waiting for that box to spew a sequence of words, born in lack of shared experience.

By the way, would you use and LLM to write a letter to a close friend? Or, to put that in our context, which is marketing communications on the Web, would you talk to the lady next to you on the table (yes, ref. Good old Ogily) using ChatGPT?

I wouldn’t.

I’m still reluctant to introduce in my language and thinking work the use of words in the form of tokens. When I start thinking about a text and its building idea and relationship blocks, I prefer to do that with my own muscles, so that I can keep them fit and ready to perceive and understand the world in and around me. I also want to use my text for true communication. 

It is the utterance that matters, I argue. The process of transformation which happens during the communication, not the number of words produced and the perfect-sounding set of calculations and words embedding in vectors (see https://www.understandingai.org/p/large-language-models-explained-with).

And, please, don’t get me wrong.  I have also looked forward to having smart assistants, easy to use automated tools for processing, and algorithmic helpers for legwork and tedious tasks. Not only that but I have also been tempted by the idea of looking at AI-generated text as part of the inevitable scenario unfolding described by Italo Calvino in Cybernetics and Ghosts (Italo Calvino. Cybernetics and Ghosts. In: The Uses of Literature. Essays. Translated by Patrick Creagh. p. 10.):

“Mankind is beginning to understand how to dismantle and reassemble the most complex and unpredictable of all its machines: language.”

And, hey, isn’t an AI-generated text also like infinite conversation?

Have you seen this: https://www.infiniteconversation.com?

So close to Borges’ the Book of Sand? 

And yet, so far away.

Why?

Because it is not dialogic.

A dialogue is not happening in-between the juggling of tokens. A dialogue starts where the other person enters and both participants engage in making sense together.

But, wait, I can hear your doubts…

Why would we ever want to aim for dialogue? Especially when it comes to marketing communication.

Dialogue, and dialogic orientation helps us create value, helping our messages linger in the emotional memory of the people we want to reach out to. And not only that, but the process serves us as a driving force for future rich and relevant communication. 

Choosing to generate a text in less than a minute  is popular. It saves time, and also alleviates the cognitive, and sometimes existential load when writing. Sometimes, I imagine, you need to get to a point quickly, to speed up your travel along the information highways and deliver fast. But in general, why through out the process?

Doesn’t generating a text in a blink of an eye look like taking a taxi for a walk in the park, in this case a walk in the language meanders of our shared experiences?

I don’t have a ready answer for you. And I shouldn’t have. It is up to you to choose.

I personally believe texts are not only artifacts but also processes in themselves. And there is a caveat. On the Web, for many reasons (interconnectedness and machine-readability being two of the major ones) are both an artifact and a never-ending process of sense-making.  In the state in which they exist as the latter, they have stages that you shouldn’t speed up, skip or scale.

Whatever you do, whatever you choose to write with and by, don’t fixate on text as a product only. Learn to fixate and understand text as a process.  Think about words as actions. Think about writing from the point of view of the transformation you want to see unfolding when you offer a different perspective. When in doubt about perspective, uniqness and taking the road less travelled, I always return to this scene: Dead Poets We Must Always Look At Things Differently.

And if that different perspective can come from an interaction with a large language model, fair enough. Approach text as an artifact of a certain size, with specific function and goal in mind, and skip writing as a process of connecting and co-creating value in a dialogue.

But please keep in mind that the ghost in the machine is not a deus ex machina. It is not an automagical way that will resolve what was unresolved in the socio-cultural and technical strands our texts are woven of. 

Such resolution would take the human power to use text to align views of the world, to create connections that never existed before. Connections that would enrich each of the participants’ understanding of the world around us and the world in us. 

It is that use of text that can help us go beyond the known edge of knowledge where a text becomes statistically improbable.

It is there that we would make a difference by using words to know and do more. Together. Not  All Watched Over By Machines Of Loving Grace. Yet.

p.s. Here’s the letter from a universe in whose scenario I chose to generate the letter with  ChatGPT: https://chatgpt.com/share/67b11150-bd44-8001-bb33-e85a92397048

p. p.s. All I have told you I owe to giants I am stepping on the shoulders of. People who have been threading the thread with their research. To provide a start, I’m including a list of sources I very much hope you will find interesting and helpful to empower you to choose wisely how you use tools in your knowledge work.

1. Large models of what? Mistaking engineering achievements for human linguistic agency – ScienceDirect
2. Italo Calvino: The Uses of Literature (pdf). pp. 3 to 27 
3. Roland Barthes: The Pleasure of Text (pdf) 
4. Dialogue: A Proposal 
5. AI and its consequences for the written word