New mediums for humans to complement superintelligence

If superintelligence has already been commoditized and neatly packaged into the workforce, what would our renewed “mediums of message” look like?

To understand the shape + form of medium 2.0, we might need a quick detour through the history of mediums and ‘what has been.’

We’ve experienced revolutions and centuries of evolution in how we communicate — In ancient times, communication was primarily oratory—there weren’t written instructions.

For example, the ancient Vedic scriptures of Hindu religion relied on word-of-mouth as sacred verses were passed secretly through priests. In this era, human memory capability offered an inherent advantage.

Later came the transition to textual mediums where information could be stored and documented. No longer relying on memory alone, the written form could be disseminated rapidly with the rise of the Gutenberg press, improving information reproduction, reducing errors, and helping centralize truth.

But the evolution of mediums of message didn’t stop at oration, literature, and printing. It continued to change faces. Television emerged, influencing even presidential elections. In the book titled — Amusing Ourselves to Death, Neil Postman describes how presidential debates during Abraham Lincoln’s time focused on lengthy conversations where crowds gave their undivided attention. Today, these debates have shifted toward performance, with greater attention to what’s said ‘between the lines’ and non-verbal cues — sometimes overshadowing the actual content.

Now, with social media, we exist on continuous dopamine drip-campaigns delivered through Instagram reels and TikToks, making societal discourse shorter and more attention-grabbing, emphasizing ‘hooks,’ openings, and closers. As we navigate through history’s mediums and their effects, we arrive at the present, where supposedly safe super-intelligence might eventually be uniquitous.

If that’s the case, we might have to reflect critically on the mediums of choice. I make the case that, online essays which has been a highly popular medium for sharing ideas will get rechristened into something more malleable so that we could ‘talk’ to superintelligence better.

Essays would combine with software: it’s the shape of a software with an overlapping essay to complement each other. This future is already here, it’s just that it’s unevenly distributed. The Quantum country essay by Andy Matuschak, where he interweaves essays with interactive flashcards to help retain the ideas better, are one of those first examples where I’ve seen this medium of software + essay catch internet’s attention. Tools like Cursor offer early glimpses of these translation layers, are now creating spaces where humans / AI collaborate on shared goals with different but complementary capabilities. The format of .cursorrules even acts as a replacement for PRD, to communicate requirements for ‘vibe-coding’ products—instruction formats made clear for AI agents writing code on our behalf. Or even the way in React animations are sprinkled onto technical blogs (Benjamin Dicken’s blog on IO devices is one example on this front) — essays with a hell lot of micro-interactions.

Currently, our documents speak to human understanding—developers read them, interpret them, and convert those ideas into code. But we’re entering a middle ground where AI agents begin taking on development tasks. Our documentation needs are shifting to serve both human and machine audiences simultaneously.

These documents might feature simulations that agents can run to test assumptions, rich metadata providing AI contextual understanding, and version histories tracking not just what changed but why decisions were made. Nikunj Kothari especially talks about this evolution of the beningn PRD format to now also serve prototype management purposes, in this essay he writes —“Instead of preparing for hypotheticals, you’re discussing real choices you made. Building teaches you more about a company’s market in a week than months studying their website. The conversation shifts from theoretical scenarios to actual decisions”

Our role in this post-superintelligence era might shift from mere instruction to curation of experiments. As superintelligence joins our teams, we’ll gravitate toward areas where our ‘humanness’ poses an inherent moat. The moat of what “humanness” is still evolving, but it’s safe to say that the ability to (a) understand users, (b) perform various tests to understand assumptions, experiments to validate hypothesis etc might still remain under the human moat category.

I also see some other debate-able human moats (such as taste, emergence of unique ideas, emotional intelligence, unique shared human experiences etc). We’re not rick-rolled by AI in these categories yet — we shouldn’t be.

All the more reasons why the ‘software + essay’ medium might flourish — as only a human is best positioned to empathise and understand another fellow human (or) a group of humans and their behaviors. Silicon-based meatspace (superintelligence) even in this unprepared future, might still have a long way to go.