The Beginning Is Yours: What Watching a Show Backwards Taught Me About AI

Or, how to stop prompting like a robot and start thinking like a writer.

I didn’t know who this girl was, or why she was there. I just knew she was in trouble. That was the first scene I saw in House of Cards. I had never watched the show before. And I didn’t start at the beginning. I started at the end.

In 2014, I watched the entire series backwards—one episode at a time, in reverse order—and took notes as I went. No context. No character map. No idea what I was looking for. Just this feeling that if I pulled on the string long enough, a story might form.

---

If you’re into experiments like this—on language, AI, or education—subscribe here to get future posts.

---

Why would anyone do this? Fair question. But it started with a different one: Does it really matter what order we consume stories in? If you flip the sequence, does the brain still build a narrative? Will it still need to? Will it invent structure where there is none?

That question wouldn’t let me go. And when I saw Nicholas Thompson’s January LinkedIn note, I was reminded of the experiment I ran a decade earlier. He mentioned that his friend Richard Brody—a film critic—watches movies both forwards and backwards to understand how they’re built. Hey, I thought. Me too.

Back then, House of Cards was one of the most popular shows on television. It had a reputation for being a gripping narrative. Perfect.

However, you can’t just press play and watch a show “go backwards.” So I did it manually. I started with the Season Three Finale—the most recent episode at the time—and worked my way back, one episode at a time. Watching Rachel Posner (nameless to me then) wake up in a halfway house and navigate off-the-books jobs was disorienting. I didn’t know who she was, where she was, or what she cared about.

But that’s no different from any other pilot. Watch the first episode of Severance. You’ll feel lost. That’s the point.

If you push through—and if the writing is good—you’ll be rewarded with just enough clarity to keep going. As my creative writing mentor Jeff Talarigo used to say, “There’s a fine line between mystery and confusion—you want some mystery, but you don’t want your reader to feel confused.”

So I kept watching. And taking notes. What did she do, I scrawled in my notebook. That became the objective.

Within a few episodes, a narrative began to form. Rachel and Doug clearly had a history—he kills her at the end of the Season Three Finale, after all. That was my first episode. So for me, the murder became the pilot cliffhanger.

My brain hungered for answers to the narrative gaps the same way it does when watching something forward. Why is this happening? Who are these people? What happened to them? What’s going to happen “next”?

My favorite part about that last question is that “next” is only determined by what you—the viewer—don’t know. That’s why so many stories begin at the end and then jump to flashbacks.

Pulp Fiction begins in media res. Citizen Kane starts with Kane’s death. Even The freakin’ Odyssey opens near the end. Odysseus is at a Phaecian dinner table, telling his story in retrospect—after the monsters, the storms, the Sirens.

Side note: Does anyone believe that guy Odysseus anyway? Kind of a ‘big storyteller,’ no? Sort of reminds me of a certain prominent someone…

You get the point.

In my version of the Rachel-Doug story: Girl gets chased by guy. Guy kills girl. Then, in reverse, I watch him wrestle with guilt, confusion, obsession—as I slowly gather clues about what she did to “deserve” this. My story ends when I finally understand what broke them in the first place… which, of course, is also the last place I arrive.

The actual narrative? More traditional: They fall into something like love. She betrays him. He hunts her. He can’t decide. Then he does.

So which is more powerful? Knowing the what and waiting for the why? Or knowing the why and bracing for the what?

It doesn’t matter. The point is that when you are engaged in writing—any type of writing, whether it be creative, expository, analytical, or persuasive—you are the one who decides the beginning. The construction of a thesis is quite literally a decision. It’s not set in stone. You evaluate a massive—arguably infinite—number of different pathways to your end goal before making a choice regarding where to start. When you write fiction, it doesn’t just magically start. You decide where to start.

And it’s also the job of a good prompter.

Writing is really just decision-making in disguise. So is prompting. You decide where to start, what to emphasize, what to leave out, what tone to take. If writing a story is laying a trail of breadcrumbs, prompting an LLM is doing the same—with a very strange, very powerful co-author.

---

What beginning would you choose? Leave a comment—or share this with someone who's wrestling with the same question.

---

From Transcript to Prompt

Over the last year, I’ve spent a lot of time analyzing transcripts—the back-and-forth between humans and Large Language Models—in search of patterns, blind spots, and signals for better prompting. And I keep coming back to this idea of choosing the beginning.

From what I can tell, most users bounce between three modes: productivity, brainstorming, and learning. But they don’t always know which one they’re in. I've seen users start in brainstorming mode, then realize they don’t understand the topic and shift into learning. Or they begin by asking to learn something and pivot straight into production without noticing. It’s chaotic—but fixable.

If we know these modes exist, we can teach people to pause and ask a simple question: What kind of conversation am I trying to have?

Your answer should shape your opening prompt.

Take the Learning prompt I teach in my workshops:

If I “start” the chat with an admission that I am a novice, I have immediately lowered the bar for the chat. I am no longer focused on producing something but instead seeking to gain some basic conceptual knowledge of a topic that can (maybe) guide brainstorming or productivity later on.

The user here is choosing a beginning that invites clarity instead of complexity. And that decision ripples through the rest of the conversation.

Some of this may sound obvious to well-educated adults. But that’s kind of the point. We learned this stuff before AI “existed.” We practiced it through essays, lab reports, class discussions, and teacher feedback.

Now, students are being thrown into AI-powered writing and thinking without a framework for how to engage. And what happens? They slide between modes without realizing it. They ask a learning question, then copy the output like it’s a finished product. They produce before they understand.

Unless we teach AI use through this lens.

Hence, ergo, vis-a-vis—we need to reflect before we prompt.

Where is the beginning?

What Happens When We Don’t Choose the Beginning

My fear is that we continue to engage with AI in a haphazard, arguably sloppy manner. “Just talk to it like it’s a person” is a fine starting point for someone who’s never used an LLM. But it can’t be where we stay.

I’ve seen the chaos in students. I’ve seen it in adults. I’ve seen it in myself. We need better prompting habits, and that starts with structure. Not rigid rules—just narrative awareness. A sense of order. A feel for how we enter a conversation.

The fix isn’t shaming people. It’s giving them a better route.

Analyze transcripts. Set benchmarks. Approach LLM use like a writer, not just a user.

Decide your beginning.

Because someday, we’ll look back at this early stage of prompting and realize we were just trying to keep the bot on the edge of its seat.

The Beginning

This essay is a beginning of its own. Or maybe it’s the middle. It’s certainly not the end. But your answer, as the reader, depends on a red wheelbarrow. Where did you start? What experiences are you bringing? How are they impacting the way you consume it?

It’s the beginning of my formal AI and creativity research. It’s the beginning of an effort to show educators that students can demonstrate thinking in the chat. It’s the beginning of a story that I hope never ends, and contains multitudes—of people, ideas, and plot twists.

Because without a good story, what is life anyway?

Interested in exploring this with your team, faculty, or organization? I run workshops on AI literacy, prompting strategies, and transcript-based assessment. Reach out if you’d like to talk: mike@litpartners.ai