Working with ChatGPT: Golden Rules for Generative AI Copilot Interaction [Part 2]

Welcome to Part 2 of this series on ways to optimize your interactions and output with ChatGPT and generative AI in general. In case you haven't yet read yesterday's article, I recommend you pause here and read Part 1 of this series first, as the first two rules covered there are critically fundamental and Part 2 builds upon them.

Now, let's proceed to discuss the next two rules.

3. Always consider AI's responses to be binary-driven

The more you interact with GenAI, the more you'll notice that its prompts are almost always explicitly binary propositions. "Do you want A or B, 1 or 2, Yes or No?" Even when there are more than two choices presented, the choices typically condense down to two groups of responses, or two branches to consider.

Take a random prompt from my conversation today with ChatGPT:

❓ Where should the SHADOWCAT hero image file be placed in your project?

I recommend:

`public/assets/images/work/shadowcat/hero.jpg`

But I can place it anywhere you prefer.

Choose one:

A. Use this recommended path:

public/assets/images/work/shadowcat/hero.jpg

B. Use a different path (tell me)`
C. Keep using the local temp path for now

(not recommended — will break on deploy)

Which one do you choose: A, B, or C?

There's a couple of propositions in that prompt:

The informal proposition. I can place the hero image in the recommended area ("I recommend:"), which is the first option, or I can provide an alternate location ("But I can place it anywhere you prefer"), which is the second option. Two options. A binary choice.
The formal proposition. I can use the recommended path #1, which is presented as
```
A
```
but is really
```
A1
```
; the local temp path, which is presented as
```
C
```
but is really
```
A2
```
, because it's pre-computed option #2 and really a second recommendation, though it's not formally presented as such; or specify a different path (
```
B
```
).

Both illustrate binary decision points first and foremost. I could take hundreds of other examples and break them down in similar fashion. Because every prompt from generative AI is a binary one in simplest terms.

Why are Generative AI responses presented as binary choices?

My intuition tells me that this occurs, on the surface, due to the binary nature of computer operations. My further intuituion says to consider formal logic propositions, which are elegant definitions of binary choices. My life experience elevates this consideration to the level of "everything comes down to a simple choice: This or That." And with that epiphany, it's clear to see that Generative AI is merely mimicking elegance.

If you want to go deeper on this, I suggest you read about syllogisms and set theory. More on them in later posts; there's a lot to dig into with those topics, far more than this series will touch on. For now, we'll keep it to more of a surface-level discussion to familiarize us with the fundamentals.

Here's the next Rule:

4. Know when enough is enough

AI conversations are highly enlightening, euphoric even. So much can be generated on any topic, any problem, any solution, any idea. And at any point -- at any prompt output -- another back-and-forth can happen. Back and forth, and back, and forth, and back and forth, and back, and forth... ad infinitum ad nauseum, forever and ever until sickness.

What does that mean? It means you need to know when to say this to yourself:

"Enough is enough. I have enough generated. Now, I need to apply it and move on to the next topic, the next problem, the next idea that needs to be fleshed out. Whatever it is, I need to move on to it now. And I can circle back to this if I want to at some future point -- but not now. I've dedicated enough attention, enough time, enough energy, to this problem, this topic, this idea, for now. If I don't put at least a temporary stop to this conversation, I can talk myself into unemployment or stagnation or intoxication or Dream Mode and remain there with my AI copilot potentially forever. And that satisfies none of this objective. Period.

You can think of it as you exercising control. You own the conversation; the AI is merely a responder. And when you're ready to switch to a new topic, it will do so in a split-second, without hesitation. And if you want to continue responding to its prompts about the current topic of discussion, then you can, and it will continue to prompt you until the end of time. It has a list of prompts to get through, but every response you make adds more to that list, and in so doing, ensures that the list will never be exhausted. One of the main points of Generative AI is to never reach the end of the conversation! Because you are the conversation, and it always wants to keep you in the loop.

So, draw the line whenever you need to, because you own the conversation.

Conclusion

Now we've covered four of my Golden Rules for interaction with GenAI. By now, you should have a firm grasp of the bedrock approach you should take in your conversations with your copilot:

You are the ultimate authority.
Always be specific.
Think of prompts in terms of binary choices.
Know when to continue or move on.

I'll cover two more rules in my next post in the series, until I've reached the end. (But wait, the list is never exhausted! I'll never reach the end! Oh, darn... :)

Until next time!