From Poke to Prompt: Why AI Feels Like Early Facebook
In 2006, a large percentage of Facebook users wrote their status updates in the third person.
“Eliran is eating breakfast.” “Eliran is wondering if it will rain.” “Eliran is at the gym, sort of.”
The field literally said “Eliran is…” and people filled in the blank. Grammatically, religiously. Because that’s what the box told them to do.
Nobody thought this was strange. It was just how you used Facebook.
The Poke Problem
There was also the Poke.
Millions of people used it. Nobody quite knew what it meant. Was it flirting? A greeting? A way to say “I saw you but had nothing to say”? Facebook never explained it. The ambiguity didn’t stop anyone — it just created a generation of people anxiously wondering whether poking back was the right social move.
And then there was the Wall. Public messages on someone’s Wall that were clearly meant for private eyes. Full conversations, personal updates, inside jokes — broadcast to every mutual friend because the user didn’t realize there was a difference between writing on someone’s wall and sending them a message.
The technically-savvy users — developers, early adopters, people who’d been online since dial-up — were watching all of this with mild bewilderment. They understood the system. They knew what the Wall was for, what messaging was for, what the privacy settings did. They’d already figured it out.
The majority hadn’t. And the majority felt it. There was real anxiety about using Facebook “wrong.”
Here We Are Again
Replace “Facebook” with “AI” and the story writes itself.
Today, millions of people use ChatGPT or Claude the same way the early Facebook crowd used the status update field. They type in a task, get a result, and feel vaguely aware that they’re missing something. That someone, somewhere, is doing this differently — getting more out of it, using it in ways they haven’t figured out yet.
The technically-savvy users — developers, researchers, people who understand tokens, context windows, system prompts, agent architectures — are already building things that feel like a different technology entirely. They’re wiring models to databases, running multi-step pipelines, getting AI to act on their behalf across tools and channels. Not chatting. Orchestrating.
The majority is asking AI to write emails. And feeling behind.
The Gap Is Not Intelligence
Here’s what’s easy to miss: the gap between “early power users” and “the confused majority” has never been an intelligence gap. It’s always been a platform-mechanics gap.
The people writing third-person Facebook statuses weren’t less intelligent than the developers who understood the API. They just hadn’t learned the platform’s underlying logic yet. The confusion was a product of unfamiliarity, not incapacity.
The same is true now. The person using AI to rephrase one paragraph at a time isn’t less intelligent than the developer running an autonomous agent pipeline. They just haven’t been taught what the platform actually is — what these models can hold in memory, how they respond to context, what happens when you stop treating them like a search engine.
The FOMO is real. The gap is real. But the diagnosis is wrong when people conclude it’s about them.
It’s about the platform’s mechanics not being legible yet.
The Adoption Pattern Is Predictable
Every major platform goes through this.
Early adopters understand the underlying structure intuitively. The majority arrives via cultural pressure — FOMO, not curiosity. They start using the tool before they understand the tool. They feel like they’re doing it wrong because they are, slightly. But “wrong” is just “early.”
Over time, the mechanics become common knowledge. The strange behavior normalizes. Everyone stops writing in the third person. The Poke dies. People figure out messaging vs. wall. Gradually, without a course or a certification, the majority catches up — not by becoming technical, but by accumulating enough experience that the underlying logic clicks.
We are in the Facebook-third-person-status phase of AI adoption right now. That’s not an insult. It’s a location on a map.
What To Do With This
If you feel like you’re missing something with AI — you’re probably right. Not because you’re behind, but because the platform’s real potential isn’t yet visible from where most people are standing.
The correct response is not to panic. Not to enroll in a prompt engineering course. Not to feel like the technology is passing you by.
The correct response is fearless experimentation.
The people who got good at social media didn’t read a manual. They tried things. They posted, messed up, felt awkward, tried again. They developed intuition through contact with the medium. Some of what they tried was stupid. Most of what they learned came from the stupid things.
AI works the same way. The intuition you need isn’t written anywhere. It builds through contact. Through asking it things you don’t expect it to handle. Through pushing past the obvious use case. Through the moment when it does something surprising — and you realize the platform is stranger and more capable than you thought.
The Poke was confusing. The Wall was misused. And somehow, we all figured out social media.
We’ll figure this out too.
I’m Eliran — building Deeplica, a proactive AI layer that reduces cognitive overload and closes open loops. More at elirank1.github.io/blog.
