In Arrival, Dr. Louise Banks does not just learn the heptapod language. She begins to think in it.
That is what makes the movie such a strong parallel for AI.
The heptapods arrive with a language humanity does not understand. At first, the reaction is confusion, fear, fragmentation, and conflict. Then a smaller group of people chooses to do the harder thing: slow down, learn the language, understand the system, and figure out what it actually means.
That feels very familiar.
AI is not just another software tool. It is a new language layer for work, creativity, decision-making, and communication. And like any new language, it changes what becomes possible once you learn how to use it.
AI as a New Language
One of the most important ideas in Arrival is that once Louise understands the language, she does not just communicate differently. She sees differently.
I think AI works the same way.
At first, most people treat it like a novelty. They type a few questions into a model, get a response, and move on.
But real fluency is different.
Real fluency means understanding how to communicate with these systems in a way that produces useful outcomes. Increasingly, that does not just mean writing better English prompts. It means understanding the technical language that sits underneath the interaction:
- Python
- SQL
- HTML
- APIs
- Databases
- Structured knowledge
In the world of AI, code is part of the language.
The better you understand those layers, the better you can communicate with the model and the more powerful the result becomes.
The Two Levels of AI Adoption
Most organizations are at level one: using AI tools. They have ChatGPT licenses, a Copilot subscription, maybe a few automations in Zapier.
This is like learning basic phrases in a foreign language. You can order coffee. You can ask for directions. You're not thinking in the language yet.
Level two is different. Level two is when AI changes how you design systems, how you structure information, how you make decisions. When the presence of AI-compatible thinking is baked into your organizational DNA.
That's when the language starts to reshape perception.
What AI Fluency Actually Looks Like
I've seen it at the organizations we've deployed for:
At VIDA Fitness, the team didn't just get a chatbot. They restructured how they think about lead qualification — building explicit decision criteria into an AI agent that had never existed in their process before. The 25% conversion rate wasn't just an AI outcome. It was the result of AI forcing clarity about what "qualified lead" actually means.
At PenFed, the Text-to-SQL project didn't just speed up data access. It changed the analytics team's relationship to questions. Instead of "what report should we build?" the question became "what do we want to know?" Less than 1% variance against source data was the technical result. The cultural result was an organization that became more curious because curiosity got cheaper.
At Axle Informatics, a multi-week proposal process got compressed below one week. But the more significant outcome was this: the team started thinking in proposal structure. The AI scaffolding made the process explicit, and the explicit process made the thinking better.
That is what fluency looks like in practice.
It is not just using the model. It is learning how to structure information, workflows, code, and data in a way the model can actually work with.
The Geopolitical Dimension
Arrival also goes one step further. By the end of the film, the United States and China move toward cooperation because they realize this new language is too important to approach purely through fear.
That part matters.
It may be optimistic to think that real-world AI cooperation between the United States and China will naturally happen, especially given current tensions. But the movie still offers an important idea: a new intelligence layer can either become a source of conflict or a reason to communicate more effectively.
The AI governance challenge between the United States and China requires the same kind of thinking.
We are two nations developing the most consequential technology in human history with nearly zero shared governance frameworks. The analog to nuclear non-proliferation exists conceptually. It has not been built.
The Arrival parallel is instructive: the heptapods came offering a gift — a new language. The world's militaries almost responded with violence. Survival required the choice to communicate instead.
The choice about AI governance is analogous. The cooperative path requires both parties to accept that the other's perspective is not illegitimate — just different.
What This Means for You
Learning AI fluency is not just about taking courses or reading research papers, though both can help.
It is about building the ability to communicate with these systems in a deeper way.
That means:
- Understanding how prompts, code, and data work together
- Structuring workflows so AI can participate meaningfully
- Learning enough technical language to guide the model well
- Building organizations that do more than just "use tools"
The goal is not to turn everyone into an engineer.
The goal is to help more people become fluent enough to work effectively with the new language layer that AI introduces.
That's the shift from translating to thinking in the language.
Louise Banks didn't become fluent in heptapod by studying grammar. She became fluent by using the language under pressure, in high-stakes situations, until it changed how she saw time itself.
The same transformation is available to any organization willing to go through it.