Apparently being polite to AI is causing the earth to melt…

There’s been an interesting click-bait-y debate making the rounds online lately:

Should we be polite to AI tools like ChatGPT?

Some argue that including words like please or thank you in prompts is wasteful. The reasoning is that these “extra” tokens (the individual word fragments AI tools process) accumulate across billions of prompts, resulting in a measurable impact on the environment through increased energy use.

The “Please” Problem: Where the Concern Comes From

The discussion started picking up steam after articles in The Washington Post, MIT Technology Review, and Scientific American raised concerns about the growing carbon footprint of generative AI tools. AI models like GPT-4 process every input as tokens—and yes, even “please” counts. Multiplied across billions of prompts, unnecessary politeness could technically result in more compute cycles and more electricity used.

Fair enough. AI tools are not sentient; they don’t care if you’re polite (although they might… check out this Forbes article ). Being efficient in your inputs is a reasonable goal.

But focusing on manners misses the forest for the trees.

Where the Real Waste Happens: Bad Prompts

If you’ve ever used an AI tool and found yourself endlessly rewording your prompt to get something even close to what you needed, you already know this truth:

It’s not the please that wastes energy. It’s poor communication.

Let’s break this down:

• Vague prompts like “Write something about marketing” often produce generic, unhelpful responses.

• Overloaded prompts try to ask for five different things in one go and get jumbled output.

• Unstructured requests leave the AI guessing what tone, format, or audience you want.

• No context leads to bland results that miss the point entirely.

Each one of those flawed prompts usually leads to a cascade of further attempts: rephrasing, clarifying, editing, discarding, trying again. That’s dozens of queries where one well-crafted one could have done the job.

This is where the real inefficiency lies. Not in being courteous—but in being unclear.

Want to Be Efficient? Learn to Prompt with Purpose

If you want to reduce AI-generated energy consumption and get better results faster, the answer isn’t “cut the fluff.” It’s “get methodical.”

Use a framework like this when writing prompts:

1. Set the Role:

“You are a senior product manager with experience in SaaS.”

2. Define the Task:

“Write a 300-word blog post summarising the risks of launching without user testing.”

3. Clarify Output Format:

“Use a professional tone. Include a clear headline and 2–3 subheadings.”

4. Provide Context:

“This is for a LinkedIn audience of early-career product professionals.”

This structured, purposeful approach does far more to reduce wasted compute power than dropping a few politeness words ever could.

AI Mirrors Your Mindset—And Your Social Skills

Here’s another layer to this conversation that rarely gets discussed:

If you don’t know how to talk to people, you won’t know how to talk to AI.

• People who communicate clearly, with intention, and who understand their audience tend to get better results in conversation.

• The same is true when prompting AI.

If you’re used to vague, indirect, or chaotic communication with humans, you’ll likely carry that into your prompts: jumbled ideas, unclear expectations, poor tone, and zero context.

AI doesn’t read between the lines like a human might. It needs clarity. It thrives on specificity, structure, and intention. That’s why the best prompt writers tend to be:

• Strong communicators

• Emotionally intelligent

• Thoughtful about language and meaning

• Capable of thinking through audience, purpose, and structure

In short, people skills translate into prompt skills.

Improving your ability to write clearly and empathetically won’t just help you in relationships or the workplace, it will also make you far more effective with AI tools. And that translates into more relevant results, fewer queries, and yes, less wasted energy.

What Efficiency Really Looks Like

Saying please to AI isn’t making the world melt on its own. Sure, token bloat at scale is a valid concern—but it’s a drop in the ocean compared to the inefficiency of misusing these tools altogether.

If you want to use AI more sustainably:

• Learn how to prompt with structure and clarity

• Treat communication as a skill worth developing

• Save energy by reducing iteration, not by stripping away personality

Case Study: Writing Efficient AI Prompts for Product Requirements

As an example, let's look at the following scenario:

You are tasked with documenting requirements for a new feature within a mobile application that will place restaurant locations on a map. As a starting point, you want to use an AI tool to generate the initial set of requirements and highlight other considerations.

Inefficient Prompt: The “Just Wing It” Approach

“I need requirements for a map feature for restaurants?”

What happens?

This vague prompt might return a basic list of features like:

• Show restaurants on a map

• Let users click on a pin

• Display details

Sounds fine… but it’s incomplete, too generic, and lacks critical detail about user needs, platform constraints, or goals. You’ll spend time refining it manually or asking multiple follow-up prompts.

Waste generated:

• Multiple rounds of back-and-forth

• Rewrites

• Poor alignment with engineering team

• Increased AI usage = more energy

Efficient Prompt: The Structured, Intentional Approach

Prompt:

“You are a senior product manager experienced in mobile apps. Please write a concise product requirements document for a new feature in a restaurant discovery app.

This feature allows users to view restaurants on an interactive map within the app.

Requirements should include:

– Feature description

– User stories

– Functional requirements

– Non-functional considerations (e.g., performance, GPS accuracy)

Target platform: iOS and Android.

The feature is intended to increase restaurant discovery and user session time. Write it in a tone suitable for an engineering and design audience.”

AI Output (Summary)

Feature Description

Introduce a map interface within the mobile app that displays nearby restaurants as location pins. Users can tap a pin to view restaurant details, navigate to the listing, or get directions.

User Stories

• As a user, I want to view nearby restaurants on a map so I can explore options visually.

• As a user, I want to tap a pin to view restaurant details.

• As a user, I want to get directions to a restaurant using my preferred maps app.

Functional Requirements

• Display interactive map using MapKit (iOS) and Google Maps SDK (Android)

• Fetch and render restaurant locations as pins with real-time location updates

• Tapping a pin shows a preview card with name, rating, and distance

• Tap on preview card to navigate to full restaurant profile

• Button to open directions in third-party map apps (e.g., Google Maps, Apple Maps)

Non-Functional Requirements

• Map must load within 2 seconds under normal network conditions

• Pins should cluster when zoomed out

• GPS location accuracy within 20 metres

• Cache map data for improved offline experience

Results

• 1 prompt = High-quality, usable output

• Minimal editing required

• Clear communication with design and engineering

• Saves energy, saves time, and improves alignment

The difference isn’t just in what you ask, but how you ask.

A well-structured prompt mirrors how great PMs structure their thinking: clear, intentional, and user-centric. It reduces the amount of rework and time/energy wasted.

If you wouldn’t ask your tech lead or designer to “just make a map thingy,” don’t ask AI that way either.