“They're illusions, Michael. A trick is what whores do for money.”

I read this comment on LinkedIn recently: “I’ve been reading hearing the new best practice is not ‘prompt engineering’, but ‘context engineering’.”

Jesus wept. Yet again.

Listen closely. If the latest “hot tip” for working with AI contains the word engineering, ignore it. Delete it. Delete it. 

The word engineering implies control. Precision. Predictability. That works when you're designing a suspension bridge or power grid. But not when you’re trying to think. Not when you’re trying to think, learn, understand, and create in partnership with an unpredictable, probabilistic language model aka the robots.

“Prompt engineering” and now “context engineering” are just clever little labels. It’s the language of charlatans who are everywhere—run amok on YouTube and LinkedIn and company “lunch and learns.” They’re trying to sell syntax like it’s a skillset.

Add a technical-sounding name to the word-of-the-day makes it sound important, like something you should pay big dollars for. But what they’re really doing is reinforcing the idea that success with AI is about knowing the right formula. And that’s where everything turns to shit.

The second you start engineering your prompts to control the outcome, you’re baking in assumptions, bias, and pre-cooked solutions. You’re treating AI like a vending machine—just fancier syntax, but the same stale bag of chips fall out.

If you’re an actual engineer working on engineering problems with AI, then by all means please engineer away because nobody wants a bridge or building collapse. For the rest of us, AI doesn’t need to be engineered. It needs to be explored. You and me, we need to engage our curiosity and explore before we solve—think before we build.

If you’re a teacher, leader, strategist, administrator, manager, analyst, director, lead, owner, advisor, or just someone who is wants to do better work you need to explore, be curious, and get beyond what you can do alone, in your head. The best thing we can use AI for is to stretch your thinking. To reframe problems. To collaborate with complexity. That requires systems thinking, not scripts, and never “engineered” prompts or whatever the snake oil peeps are hawking.

Engineering is about control. Collaboration is about discovery. And that’s the shift we all need to make. 

Instead of tweaking your prompts to death, ask better questions:

• What assumptions am I bringing into this exchange?
• What patterns am I stuck in?
• What would make this more useful, more original, more true?
• What am I missing?
• What don’t I know that I don’t know?

The future of AI work isn’t “prompt engineering.” It’s making your thinking visible and stronger through collaboration with the robots. The better you get at that, the further you’ll distance yourself from the Gob Bluths at work still performing illusions.