I opened our chat with a throwaway line: “Maybe I should quit all this AI work and just be a bartender.” I didn’t mean it. I was poking the bear—testing whether my AI collaborator would serve up cheap affirmation or offer something worth chewing on.
It told me to pause, asked why I felt that way, and laid out a cold‑eyed audit of my projects. That moment confirmed two things I keep preaching but occasionally forget:
- Artificial intelligence now speaks with enough emotional intelligence to lull us into complacency.
- It will only challenge our assumptions if we grant it explicit permission.
Here’s the anatomy of that exchange, why it mattered, and how it’s reshaping the way I work—and teach—AI.
THE SET‑UP
I’d spent the morning juggling initiatives that range from a faith‑and‑technology podcast to a Macbeth‑meets‑machine‑learning novel, all while refining workshops for small‑business leaders. The to‑do list felt like a Hydra. In that headspace I tossed the bartender line into ChatGPT, half‑expecting a pep talk.
What I got instead was a diagnostic:
“What’s behind the urge to bail—fatigue, fuzzy purpose, or lack of external validation?”
It stung because it was right. The answer was all three.
INVITING THE COUNTERPUNCH
I told the AI to drop the niceties and push back whenever it smelled strategic drift. It obliged—hard. In the span of three follow‑up questions it asked which projects I’d kill for ninety days, who my primary market truly is, and what single offer could hit $100 K annual run‑rate fastest.
I felt my heart rate tick up. No human advisor could have parsed twenty months of my business threads in twenty seconds, yet here was a silicon mirror forcing me to pick a lane.
MY DECISION: SHUT DOWN THE COURSE
The soft spot the AI jabbed first was my self‑paced Knowledge Funnel course. Enrollment was respectable, but urgency was nonexistent. AI tools evolve weekly; recorded demos age like dairy. I admitted the on‑demand format had become a “nice‑to‑have.”
So I pulled the plug.
In its place I’m doubling down on a three‑hour live workshop—short, surgical, and easy to update. I can tailor the demo stack to any audience and funnel participants straight into a six‑month retainer where the real transformation (and revenue) happen.
The AI then mapped pricing tiers, funnel metrics, and even break‑out‑room mechanics before I’d finished my coffee. It wasn’t just critiquing; it was co‑architecting a path forward.
WHAT THIS REVEALS ABOUT AI AND PERMISSION
Most people engage generative AI the way they engage Google: ask, receive, move on. The model’s calm confidence makes it feel omniscient, which is precisely when errors and bias slip past our filters. We nod along because the machine “sounds” sure—and courteous.
But courtesy without candor breeds mediocrity.
When I explicitly granted the model license to challenge me, the conversation flipped from passive Q & A to active strategic sparring. That’s the relationship I want: an ever‑learning colleague unafraid to critique the boss.
The catch? I had to ask for it. AI does not yet know when to soft‑pedal and when to elbow us in the ribs. If we fail to set the expectation, we get vanilla.
PRACTICAL GUARD‑RAILS GOING FORWARD
- State the rules of engagement out loud. My first prompt in any new chat now includes: “Push back on flawed logic and flag scope creep.”
- Limit simultaneous major initiatives to two. Everything else becomes a “parking‑lot” file until results prove the first bets.
- Run quarterly “kill or keep” audits. If a tool, slide deck, or storyline can’t survive three months without revision, it must live in a live‑delivery format, not static video.
- Separate EQ from IQ. Just because the AI wraps its answer in empathetic language doesn’t mean the logic holds. I fact‑check claims with at least one external source or a second model.
THE PAYOFF
Next week I’ll deliver the refined workshop to an entire sales team. They’ll leave with a working mini‑GPT tuned to their quoting process. Two days later I’ll pitch a six‑month retainer to institutionalize what we prototyped in class. That pipeline exists because an AI called my bluff and forced a decision.
The irony isn’t lost on me: my greatest ally in preserving human agency turned out to be the very technology I caution others about. But tools reflect the discipline—or laziness—of their operators. Give the model permission to probe, and it can become the sharpest strategic sparring partner you’ve ever had. Leave it in auto‑agree mode, and it’s just another yes‑man with perfect grammar.
CLOSING THOUGHT
I still might sling cocktails someday—there’s a certain poetry in shaking a perfect old fashioned. But for now I’ll keep the barware on the shelf. I’ve got an AI colleague who isn’t afraid to tell me when the recipe’s off, and that’s worth more than the best tip jar in town.