The Prompt Paradox
What’s the value of words, strung together, in a sentence? Absolutely worthless and valuable—at the same time.
How much are your thoughts worth?
Some say a penny. Others say a civilization.
A prompt is like a distilled and expressed thought that, in turn, leads to more thoughts. They resemble potions and spells in an old wizard's shop, or prayers offered for divine favor. We quantify them, paying freelance writers by the word—or pay OpenAI a few cents per token.
And like all language:
Prompts are both worthless and valuable, at the same time.
They’re worthless because they’re just words in the form of commands, statements, or questions that don’t require you to get the wording just right.
You can prompt ChatGPT for the same thing, twenty different ways, and get useful output that’s very similar. You do not have to get the words just right as long as you know your intention and what you’re looking for.
At the same time, prompts are valuable because knowing what’s possible with prompting can sometimes help you eliminate wasted time and effort through trial-and-error.
Because it’s all language and words, prompts are also infinite in variation.
There are a hundred different ways (or more) that you can have ChatGPT produce anything from marketing strategies to Google ads, and everything in between.
So, they’re worthless on their own. Because there’s no one single formulation of a prompt that will always produce the same output, every single time.
But they’re valuable in that if you learn how to structure a good prompt, and understand that you’re engaging in a conversation (a chat) to get better output—you can consistently get great output.
What makes a prompt worthless?
Language is nearly infinite. You can express the same message in multiple ways. Therefore, prompts are also endless.
What matters is how clear your intent is, not necessarily the exact words.
Other reasons why there’s not a perfect combination of exact words that will always give you the best output:
You can use the same exact prompt ten times in a row, and get variations, nuances, and differences in the output ten times. You’ll never get the same exact output. They’re probabilistic and non-deterministic.
Why?
Because LLMs are narcissistic.
They “prefer” their own words over yours. And never, ever follow the exact letters and words in your prompt.
But LLMs follow the direction you give, in the intent.
They turn your words into numbers and never preserve the exactness of your words.
Therefore, there’s not a single perfect prompt anywhere. It’s not even technically possible because of how LLMs treat input and output. What matters is the intent and direction of your words.
It’s the same as with humans and language. If a human does not understand you the first time, you change what you said but preserve the intent.
At the same time, prompts are valuable.
What makes a prompt valuable?
In everyday communication, we use prompts to start or continue a dialogue or command an action.
A prompt can be a question, a request, or a simple instruction. Its role? To elicit a response, whether from a machine or another person.
At first glance, a prompt is just a collection of words. But its power lies in its ability to guide the conversation or interaction towards a specific outcome.
Speech Act Theory takes this understanding further. Developed by philosophers like John L. Austin and John Searle, it suggests that when we speak, we do not just share information—we perform actions.
You’ll see a clear difference in responses from an LLM when the prompt is precise and clear versus vague or general.
The former can lead the model to generate insights, creative ideas, or solutions that go beyond the surface level.
In human interactions, a well-constructed prompt can open up chains of thought, previously unexplored, enriching the conversation and deepening the connection between individuals.
A prompt does more than convey a message—it acts. It can request, instruct, or even persuade. Speech Act Theory highlights the foundational role of language in shaping our interactions, not just with each other but also with LLMs.
When we craft a prompt for a large language model like GPT-4, we're not just inputting data. We're engaging in an act of communication, aiming to guide the model towards a desired response.
The same goes for human interactions. The words we choose, the way we structure our sentences—all of it performs an action intended to elicit a specific response.
This dual function of prompts—as conveyors of information and as actions in their own right—reveals the complexity of our interactions, both digital and personal.
The simple act of crafting a prompt involves a deep understanding of language's power to shape outcomes.
Whether we're interacting with a machine that processes our words through algorithms, or a fellow human who interprets them through a shared understanding of language, the essence of the prompt remains the same.
It's a technology and a vehicle for action.
The value of prompts lies in their ability to shape outcomes, engage thought, and foster meaningful interactions. They become keys to unlocking the potential of our conversations and our interactions with AI.
To make prompts more valuable, improve how you formulate thoughts into words. You transform the prompt from a simple string of words into a technology for creativity, insight, and connection.
The true value of prompts becomes clear when we see them as more than just questions or commands. They are the craft of language, shaped to guide thought and uncover insights.
Prompts are “Affective Technologies”
Prompts act as “affective technology”, tools that can touch our emotions and shape our thinking. This ability to influence both emotional and cognitive states places prompts at the heart of affective technology in communication—with LLMs or between humans.
Speech Act Theory again helps us grasp this value. It teaches us that with prompts, we do more than share information—we perform actions.
When we request or instruct through a prompt, we're not merely seeking an answer. We're inviting the responder into a specific kind of engagement. Each prompt carries the potential to not only get a response but to trigger a particular kind of thinking or dialogue.
Prompts move in the interplay between language, emotion, and cognition.
But don’t confuse form with function. Just because the form (what it looks like) is a prompt, doesn’t mean that the magic is in the form (the prompt).
The prompt is not magic. But the function of the prompt is.
What is a function? The intent behind the words. This is the signal amongst the noise.
The stronger the signal, the clearer the communication.
You can have the same signal expressed in different ways. Prompts are just expressions in words, of an intent. What matters is the intent.
Only intention is preserved, because by the time you send off a prompt, it’s reworked and reinterpreted by the time it hits the LLM, and then in the response.
This is why a single prompt will give you different outputs.
What you need to become good at is understanding what makes a good question, what makes a good direction, and how to communicate your intent.
Prompts are simple instructions in natural language, the way you would talk to a human.
From this principle, you can extrapolate and say:
Good prompting is good communication. Bad prompting is bad communication. If a human can’t understand your set of instructions, then an LLM will struggle with it too.
In other words: Talk to an LLM like you’d talk to a human being.
This is partially why ChatGPt took off like it did. It made the interaction with an LLM a chat session, the way you’d chat with or text a friend.
This is another 80/20 principle of prompting:
You will always get better output if you engage in a conversation with an LLM, versus going in “cold” and firing off a single prompt.
And, with Projects (Claude) or GPTs (ChatGPT), you can provide enough context (data and information) along with custom instructions, so that these LLMs can emulate any language scenario you can imagine.
Talk soon,
Samuel Woods
The Bionic Writer