When to Say No to AI (And When You're Making a Mistake)
We're living in an era where the answer to every problem is starting to sound like, "Let me ask AI about that."
And sometimes, that's exactly right. But other times, it's the worst possible move you could make.
The issue is that most people are asking the wrong question. They're asking, "Can I use AI for this?" when they should be asking, "Should I use AI for this?" And those are very different questions.
One gets you an answer. The other gets you wisdom.
THE PROBLEM WITH AUTOMATING EVERYTHING
I watched a founder automate his customer email responses using AI last year. Smart guy. Efficient system. He saved four hours a week.
He also lost three of his biggest clients within two months.
Why? Because the emails came back perfect in structure, but hollow in humanity. His clients felt replaced. They felt like they were talking to a system, not a person. The emails were warm enough to feel uncanny, but not warm enough to feel real.
He could have used AI. But he shouldn't have.
This is happening everywhere. We're optimizing for speed and scale, and losing something critical in the process: the thing that made us valuable in the first place.
The skill isn't knowing how to use AI. The skill is knowing when not to use it.
THE THREE THINGS YOU SHOULD NEVER AUTOMATE
First: moments that build trust. Any interaction where someone is deciding whether to believe you, trust you, or invest in you - that's a human moment. Your client's first message to you. Your employee's first day. The moment someone decides whether to hire you or your competitor. These need a human, specifically you or someone on your team who cares.
AI can draft these. AI can speed up your response. But the final touch, the presence, the attention - that has to be human. Anything else is visible to the person on the other end, and it feels like rejection disguised as efficiency.
Second: moments that require judgment and context. When a customer complains, they don't need an optimized response. They need someone who actually heard their complaint, who understands their situation, and who makes a judgment call about what's right. AI can flag which complaints are urgent. But the response? That needs a human.
Judgment is hard to scale. It requires knowing your customer, your values, and your willingness to absorb costs to do right by someone. No AI system will ever be as good at this as a human who actually cares.
Third: moments where the value is in the conversation, not the outcome. If you're mentoring someone, training them, building a relationship - the value is in the exchange itself. A mentee doesn't need your perfectly optimized advice. They need to see how you think. They need to feel heard. They need to know you spent real time with them.
AI can give them better advice, faster. But it can't give them what mentoring actually is: the experience of being known by someone with more experience.
THE FOUR QUESTIONS TO ASK BEFORE YOU AUTOMATE
Before you hand a task to AI, ask yourself these questions:
One: Is the human element the value, or the efficiency? If someone is paying you for how fast you deliver, automate. If they're paying you for judgment, wisdom, or relationship, don't. Simple as that.
Two: What gets lost if I automate this? Not what gets gained - what gets lost. What quality, what texture, what feeling disappears? Is that loss acceptable? Sometimes yes. Sometimes no.
Three: Could I automate this but still do it human? You can use AI to draft something, then spend 10 minutes making it real. You can use AI to speed up your research, then spend time with the person to understand their actual problem. Use AI as a tool, not a replacement.
Four: Am I automating this to serve my customer, or to serve myself? Be honest. If you're automating something because it makes your life easier, and it makes your customer feel less valued - you're making a choice to prioritize your time over their experience. That's a valid choice sometimes. But know what you're choosing.
THE JOBS THAT SHOULDN'T BE AUTOMATED YET
Here's the list of things I see people automating that make me cringe:
- First responses to new customer inquiries
- Feedback on someone's creative work
- Rejection emails (especially for job applications or partnerships)
- Condolences or apologies
- Coaching or mentoring conversations
- Decisions about firing or major team changes
- Sales calls where the relationship is new
These things CAN be automated. But they shouldn't be.
Some people use AI to draft these things and then put a human touch on top - that works. But many people just hit send on the AI-generated version and hope no one notices.
Everyone notices.
THE BOUNDARY YOU NEED TO SET
Here's what I tell people: Automate the things that buy you time to be more human.
Automate the research so you have time for a real conversation. Automate the scheduling so you're not stuck in logistics. Automate the data entry so you have mental space to think about strategy. Automate the routine so you can show up fully for the exceptions.
Don't automate the things that are supposed to make you valuable. Don't automate the relationship. Don't automate the attention. Don't automate the judgment.
The measure of wisdom in the AI age is not how much you've automated. It's how clearly you've drawn the line between what machines should do and what humans should never give up.
Say no to AI. Often. Loudly. Specifically. That's what will actually make you competitive.
Learn to use AI wisely. The future belongs to people who can use AI without losing their humanity. We teach that balance.