Zero to AI — Episode 7: The Human Side of AI Workflows
Let’s be honest for a second. How many times this week have you read an email, a LinkedIn post, or a project update and thought… “A human being definitely did not write this”?
It is that specific flavor of “perfect.” It is the overuse of the word “delve.” It is the rich “tapestry” of content. It is that hollow, cheerful, “I hope this email finds you well” energy that makes you want to throw your laptop out the window.
We talk a lot about tools in the AI space. But today, we need to talk about what sits underneath the tools. We need to talk about mid-career reinvention and building a life that still feels human, even when half your work is being done by a server farm in California.
This is the quiet crisis of the modern worker. It isn’t about how to prompt better. It is about the deeper questions:
How do I keep this personal when AI is doing the heavy lifting?
Where do ethics actually live when I’m tired at 11 PM?
How do I keep my voice, my taste, and my creative control when the machine can generate ten versions in ten seconds?
If you have ever looked at an AI output and thought, “This looks technically good… but it definitely does not feel like me,” then this post is for you.
1. The “Drift”: Why the Human Side Matters
Here is the trap: AI is seductive because it removes friction. It fills the blank page. It makes you fast. But there is a hidden cost I call The Drift.
I once asked AI to help me rewrite an email to a potential client. My original draft was a mess—I was apologetic, rambling, and insecure. The AI version came back flawless. It was confident. It was “polished and corporate.”
I looked at it and had a sinking feeling. This is impressive… and if I send it, they will have no idea who I am.
It was technically excellent, but it was a lie. If they hired the person who wrote that email, they would be hiring a stranger.
That is the risk. The more you automate, the easier it is to drift away from yourself. On the surface, things look smoother. You have fewer blank pages. But underneath, the quiet questions creep in: Is this really my judgment? Is this how I want to show up in the world?
Especially in mid-career, you have spent twenty years building a reputation and a voice. Are you willing to trade that for speed?
2. Ethics: Don’t Feed the Machine Your Secrets
We tend to think of AI ethics as big, philosophical debates about “bias” and “alignment.” But in your daily workflow, ethics happens in tiny, quiet moments.
It happens when you paste a client email into ChatGPT to draft a reply. It happens when you upload a contract to get a summary. Each one of those clicks touches privacy, consent, and power.
I learned this the hard way. I once pasted a chunk of real client text into an AI chat—names, context, personal details, the drama, everything. The reply was helpful, but I immediately felt a cold sweat. I did not need to share this much.
That moment changed how I work. Now, I use a simple Four Question Ethics Check before I hit enter:
Necessity: Should I use this at all? Am I using AI because I need to, or just because I’m avoiding a difficult conversation?
Anonymity: Can I strip the identity out?
Power: Who holds the power here?
Ownership: Will I stand behind this text if it goes wrong?
The “Lazy Prompt” vs. The Ethical Prompt
Here is a real-world example of what this looks like in the prompt box.
The Less Ethical Prompt (aka The Lazy Prompt):
“Write a firm email to [Client Name] at [Company] telling them they are late on the invoice.”
This exposes the client, exposes the company, and delegates your authority to the machine.
The Better Prompt:
“Help me write a fair, clear email about a late invoice. Keep it calm and professional. No names. Just the tone.”
Same help. Almost no exposure. Set a hard boundary for yourself: “I anonymise client information before using AI.”
3. Empathy: The “Anti-Robot” Protocol
AI optimizes for efficiency. Humans value connection. Those two things are usually opposites.
The problem is that AI doesn’t have bad days. It doesn’t get stressed. So when you ask it to write a message, it defaults to a “friendly robot” tone.
I recently asked AI to rewrite a difficult message to a colleague who was overwhelmed and behind on a project. The AI version was clean, polite, and absolutely cold. It wrote: “Please provide these by Friday so we can avoid delays.”
Technically correct. But if I received that while I was drowning in work, I would want to quit.
So I added one human line. Just one sentence:
“Hey, I know you have a lot on your plate right now, so this is not pressure, just clarity.”
That one sentence changed everything. The AI provided the clarity; I provided the safety.
This brings me to the One Person Rule. Before you prompt, stop and ask: “Who is this really for?” Don’t just say “Write an email.” Say: “Help me write a supportive update for someone under pressure.”
Empathy isn’t about writing more words. It’s about writing the right words.
4. Creative Control: You Are Not the Typist
This is where most people give up. They try to use AI for writing, it sounds generic, and they quit.
To fix this, you have to adopt a new mindset: You are not the typist. You are the Director.
A film director doesn’t hold the boom mic. But they do decide the framing, the lighting, and the final cut. That is your job now. Here is how you take control.
Tactic 1: The “Three Angles” Approach
Instead of saying “Write a blog post about X” (which guarantees a boring result), try this:
“Give me three angles on this topic. I will choose the one that feels like my voice.”
Make the AI do the drafting, but you make the creative decision.
Tactic 2: The Style Guide
You can teach the AI to mimic you. Build a simple “Zero to AI” style guide. Paste three samples of your best writing into the chat and say: “Describe my tone. Create a short style guide based on this.” Then, use that guide to critique the AI’s work.
The Result:
Default AI Voice: “In today’s fast changing world, leveraging AI is essential.”
Your Voice (with Style Guide): “The world is changing fast. You do not need AI to impress anyone. You need it to stop your work swallowing your life.”
You also need red lines. My personal red lines: “I do not publish anything I have not read fully” and “AI does not write my personal stories.”
5. Designing Human-Centred AI Rituals
If you want to keep your work human, you need rituals. Here are three simple ones I use every day:
The Intent Check: Before you open ChatGPT, ask: “What do I want this to feel like for the other person?” If you skip this, you get the robot voice.
The Source Acknowledgement: Be honest with yourself. “Drafted with AI, edited by me.” It keeps you humble.
The Human Pass: Make your final review about honesty, not just typos. Ask yourself: Does this sound like a real person? Is anything manipulative or hollow? Is there one sentence I can add to make this more human?
The Human Workflow Lab: Your Homework
I want you to leave this post with something actionable. Here are three micro-experiments. Choose just one to try this week.
Experiment 1: Ethics in 5 Minutes. Write down two “Non-Negotiables” for your AI use. Maybe it’s “No client names.” Maybe it’s “No automated apologies.” Write them down. Do you feel clearer?
Experiment 2: Empathy on One Message. Draft a tough message with AI. Then ask: “What would this feel like if I read it on a bad day?” Add one human sentence to fix it.
Experiment 3: The Tiny Style Guide. Paste some of your writing and ask the AI: “Create a short style guide for my voice.” See if it actually understands you.
If you try the AI version and it actually sounds better than you? That is okay. That is data too. Maybe that is a task you should automate.
Here is the bottom line: AI can make you faster. It can make you smarter. But the real question is whether it can help you become more you.
If you try any of these experiments, message me on LinkedIn or visit changeable.co.nz for more tools.
