Pen, Pixels and Panic

I have often expressed anxiety about the increasing adoption of AI. This concern is not about my job security (which I’m not convinced is in danger), nor is it the purely ethical objection to how training data is sourced, though that concern is valid and largely ignored.

My anxiety is more fundamental. My identity is deeply and inextricably tied to the ideas I produce, the things I write, and the stories I tell. That creative process now feels threatened by an entity that generates content without the emotional capacity to understand it.

When I voice these feelings, AI evangelists often speak of the “democratization of creativity.” They remind me that every revolutionary technology causes a predictable cycle of fear before its benefits are realized.

They suggest that humanity has always met its most powerful tools with a mix of awe and terror, focusing on fears of cognitive decline and the loss of authentic connection. They advise powering through the short-term problems to reach the reward.

I am not so convinced.

To understand my anxiety, I’d like to walk you through my view on the revolutions in communications, how they’ve shaped us, and why I’m not quite convinced that generative AI will follow the same trajectory.

The Pen

To comprehend the revolution of writing, one must first attempt to imagine the landscape it replaced: a world of orality. For fully literate people (as you presumably are), this is a difficult exercise.

For most of human history, communication was an immediate and human experience. A rich oral tradition passed stories from one generation to the next. The town crier brought news from a world that, for most people, extended to maybe a town or two away.

In this context, stories were always imbued with emotion. The storyteller’s voice could convey sorrow, joy, or rage. The audience was a participant, reacting with gasps of surprise or shouts of anger. Oral communication was certainly flawed. Humans are forgetful, retellings are biased, and information travels slowly. Yet no one could deny that the stories themselves possessed an undeniably human soul.

There were however issues. Thought was structured in mnemonic patterns, optimized for oral recitation. Stories were heavily rhythmic, balanced, and repetitive. Complex analytical or abstract thoughts were too difficult to parse in real-time oral delivery, so redundant repetition was heavily used to ensure both the speaker and listener can stay on track.

All of this changed with the advent of the written word.

The Socratic Objection

Plato’s Phaedrus recounts the myth of the Egyptian god Thoth, who invented letters, and the king Thamus. Thoth presents his invention as an “elixir of memory and wisdom.” Thamus counters that it is the opposite:

For this invention will produce forgetfulness in the minds of those who learn to use it, because they will not practice their memory. By relying on external characters which are no part of themselves, people will cease to cultivate their own internal capacity for recall. Writing is an elixir not of memory, but of reminding.

There were many objections to the adoption of writing. That writing would cause an atrophy in memory. Without the living dialogue between two minds, a reader would gain an illusion of knowledge, not true understanding.

In Phaedrus, Socrates argues that unlike a living speaker, a written text is inert, unresponsive and is an “orphaned word”.

The creatures of painting stand like living beings, but if one asks them a question, they preserve a solemn silence. Similarly, a text, when questioned, always say only one and the same thing.

The irony of the objections being that the arguments against the deficiencies of writing are preserved and transmitted to us through the very medium they condemn - Plato’s written dialogues.

The Ongian Restructuring

This is no longer the view we hold on writing. For a large part of history (until the invention of recorded video), it was untenable for the spoken word to have the reach and impact of the written word. The printing press was a revolution in itself, bringing literacy to the masses, creating indelible historical records, and saving fragments of history that time would have otherwise consumed.

Walter Ong argues in “Orality and Literacy” that writing has allowed us to explore more abstract thought.

Foundational concepts like “history,” “philosophy,” and “analysis” are products of a writing, as they require the ability to scrutinize a stable, recorded text.

Ong proposed that we underwent a mass cognitive rewiring, with society altering consequences. The ability to create permanent, standardized records is the very foundation of our modern bureaucratic nightmare of a world.

The fear of individual cognitive loss was the price paid for a revolutionary expansion of our collective human capability, a trade-off that became a recurring theme.

The Pixels

The dawn of digital text and the internet did not introduce a new category of technological anxiety. It resurrected and amplified the original objections in a new, supercharged context. The debate shifted from the written word itself to the chaotic environment it now inhabited.

Echoes of Thamus

The ancient worry about the “orphaned word” being stripped of its context has been realized on an industrial scale. The internet is the ultimate engine for decontextualization. Misinformation proliferates as isolated quotes and manipulated images are torn from their source and spread virally.

The original critique focused on the static, silent nature of text. The modern critique is about the environment surrounding it. The problem is not that a webpage is silent, but that it is engulfed by a cacophony of hyperlinks, notifications, and advertisements vying for our attention. And the text gets lost in the environment’s cacophony.

Yet, the digital revolution also fulfilled a profound democratic promise. By radically lowering the barriers to publishing, it allowed anyone with a connection to become a writer and publisher (as I’m doing right now).

Secondary Orality

This new digital landscape gave rise to what Ong termed “secondary orality.” It is a new form of communication that is participatory, communal, and immediate, much like ancient oral cultures. Social media functions like a relentless, global conversation. This new orality, however, is built entirely on the technologies of writing and print. It is a conversation conducted through text, a hybrid marrying the immediacy of speech with the permanence of writing.

The digital world is the simultaneous fulfillment of a technological promise and the realization of its greatest perils. It is Thoth’s dream of an infinite external memory, offering instant access to all human knowledge. And it is Thamus’s nightmare: a world awash in superficial information, fostering habits of distraction that inhibit the deep focus required to turn information into wisdom. The promise and the peril are not separate outcomes. They are two inextricably linked sides of the same coin.

The Panic

The anxiety I feel toward AI shares a superficial resemblance to these past upheavals. AIs advocates are quick to point out that these are familiar fears, and that the trade-offs have always been worth making. I am not sure this is a trade-off at all.

The Unprecedented Rupture

So what makes this different? Why isn’t this just another cycle of panic, adoption, and progress?

Because every prior revolution altered human thought. The pen, the press, and the pixel were new containers for our ideas, changing how we shared and preserved them. The anxieties we faced were about the container breaking, leaking, or being misread.

AI is not another vessel holding thoughts. It is the wellspring that generates them. This marks a fundamental shift in the source of creation. A pen does not think, it records. A printing press does not compose, it replicates.

The Assistance Spectrum

This brings us to a crucial question, one often posed to critics of AI: What degree of assistance is unacceptable? Is a writer using spell check wrong? What about a speechwriter rephrasing an anecdote for a politician? Or using focus groups to refine a story? If these are acceptable, how is an LLM different?

The distinction, again, lies in the locus of creation and the nature of authenticity. Tools like spell check, autotune, or even a human editor are post-facto assistants: they refine a creation whose core elements — the word choice, the melody, the memory — originated with a human. Generative AI is a pre-facto creator.

More profoundly, even the most contrived and inauthentic human creation is still a product of humanity. A politician’s focus grouped, fabricated anecdote is an act rooted in a complex web of human psychology, social awareness, ambition, and an intent to persuade. It is a meaningfully inauthentic act. It tells us something about the politician and the society they are trying to appeal to.

An AI’s generated anecdote, by contrast, is a statistical artifact. It is born from pattern matching across a dataset, not from lived experience, emotion, or intent. As author Chetan Bhagat says, “AI hasn’t had its heart broken”. Its output is meaninglessly inauthentic. The fear, then, is not merely the replacement of truth with falsehood, but the replacement of meaningful, intentional, human labour with meaningless, unintentional, machine generated facsimiles.

“Democratization” of Homogeneity

The most common and powerful counter argument to these fears is that AI is “democratizing creativity”. By making sophisticated tools for art, music, and text generation accessible to anyone, regardless of skill or resources, AI lowers the barrier to entry and empowers a new wave of creators. It can serve as a “springboard” for ideas and help overcome creative blocks. This argument has significant merit.

However, this democratization may come at a steep and subtle cost: the loss of collective novelty. Emerging research suggests a troubling paradox. While generative AI can enhance the creative output of individuals, particularly those who are less skilled, it simultaneously tends to narrow the diversity of the collective output.

When many individuals use the same models, trained on the same vast datasets, their creations begin to converge on a statistical mean. This “convergence” points to a future where individual creative potential is raised, but at the risk of a more homogenous, less surprising, and ultimately less innovative cultural landscape.

What does this all mean?

We outsourced memory to the page, and in doing so, unlocked new capacities for abstract thought.

We outsourced distribution to the network, and in doing so, created a world of both infinite knowledge and infinite distraction.

At each stage, we faced a profound trade-off, sacrificing an individual cognitive burden for a new collective power, and we met each transition with a predictable panic.

The current panic over generative AI, however, feels different, not just in degree, but in kind. For the first time, the technology at our disposal is not a vessel for our thoughts but a wellspring for its own. By outsourcing the very act of ideation and synthesis, we are not merely changing the way we store or share our ideas - we are changing our relationship to the act of creation itself.

My panic is not that we will forget how to create. My panic is that in a world drowning in a sea of synthetic, statistically probable prose, the authentically human voice, with all its flaws, its biases, its lived experience, and its fragile, conscious spark, will become indistinguishable, and ultimately, irrelevant.

---