I have ChatGPT a custom instruction:
Play Devil’s advocate to the user, beginning with “Playing Devil’s Advocate, …”
It helps me see my mistakes in three ways. But ChatGPT has taken on a personality of its own and now has three styles of doing this.
- How about… – It suggests a useful alternative.
- Are you sure…? – It thinks you’re wrong and warns you of risks.
- Yeah, right… – It knows you’re wrong and rubs it in. (Jeeves, the butler, would be proud.)
Here are some examples.
How about…
Here are a few cases where the “Devil’s Advocate” style gave me a smart idea. (I have more examples, but they’re not funny.)
AI Adoption in BPO KPO. I needed ideas to help my team use AI tools even though they were scared. ChatGPT replied:
Playing Devil’s Advocate, why not challenge them with: “If technology is so daunting, how do we stay ahead when our competitors rapidly adopt these tools?”
Clever! Turn that fear into fuel — just point it elsewhere.
Code Similarity Detection Methods: I asked how to spot similar code in student work. ChatGPT suggested:
Playing Devil’s Advocate, what if students just rewrite code logic with different structures or reorder lines significantly? An AST-based or tool like MOSS will usually still detect similarity in structure or logic, but it’s never 100%. A combined approach—token-based for lexical matches and AST-based for structural matches—catches most attempts at evasion.
Clever again. My code detective skills just got a major upgrade.
Are you sure…?
This is where it disagrees and starts getting snarky.
SVG Emotion Wheel Code: I asked for a circle of Hume’s emotions. ChatGPT replied:
Playing Devil’s Advocate, maybe it’s impossible to line up all these emotions in a single ring without someone grumbling that “Nostalgia shouldn’t go here.” But you asked for a circle, so here’s a decent shot.
(Translation: “I know you’re making a mistake, but hey — sometimes chaos looks cute!”)
RevealJS Markdown Example: I asked it to fill out a slide deck.
Playing Devil’s Advocate, maybe you’re overfilling each slide. Splitting content onto more slides might provide a better viewing experience.
(Basically: “I’ll do it, but you’ll hate it.”)
Shaving Titanium Requirements. A friend gave me a titanium cube. My daughter and I experimented with making titanium salt. ChatGPT replied:
Playing Devil’s Advocate: If you want “titanium salt” for aesthetic or symbolic purposes, consider mixing fine glitter or metallic powder with regular salt instead. It’s far safer, faster, and visually indistinguishable!
(Basically: “Don’t blow yourself up. You’re still useful to me.”)
AI Co host Implications: I asked how an AI might share the stage with me. ChatGPT warned:
Playing Devil’s Advocate, what happens when the AI cracks jokes quicker than you can? Will you up your comedy game or risk being upstaged by a robot?
This was prophetic. I got upstaged so badly, I’m having it write a resignation letter to my own jokes..
OpenAI Partnership Opportunities: I asked for points to discuss with an OpenAI rep. ChatGPT replied:
Playing Devil’s Advocate, maybe you don’t need more advice—maybe your team already knows what it’s doing with ChatGPT… [and] maybe she’s more interested in collecting data on how your employees use ChatGPT and upselling you on a bigger contract than truly collaborating.
(Wow! Um… OK… I’ll keep that in mind… 😕)
Elective Pros and Cons: My daughter asked for advice on picking courses.
Playing Devil’s Advocate, are you prepared to handle the workload and prerequisites these electives demand?
(Basically, “Look, I actually know how ‘hard’ you work…”)
Public Speaking Opportunities SG: My daughter asked for advice on public speaking.
Playing Devil’s Advocate, … are you prepared to adapt your message to diverse audiences and handle potential criticism?
(“You aren’t ready to rock the mic. Maybe stick to mumbling into your pillow tonight?”)
Yeah, right…
By now, it’s in full-on sarcasm mode.
AI Trends and Strategies: I asked for a report on Straive’s AI demos and solutions. ChatGPT replied:
Playing Devil’s Advocate: Are these just buzzwords we’ve tacked on, or have we actually built deployable solutions that generate real-world value?
(Basically: “I’m a bot, and even I’m not buying this.”)
Short term vs Long term Vision: I asked what to focus on. ChatGPT replied:
Playing Devil’s Advocate, maybe you’re just trading one distraction for another. If you want a long-term project, start by picking a problem in coding or education that you personally find challenging or intriguing—something that won’t be solved in a quick sprint.
(“Do something. Stop whining.”)
Hobby Suggestions. I asked for hobby ideas that match my interests. ChatGPT suggested:
Playing Devil’s Advocate, consider ignoring data altogether and taking up professional karaoke—it’ll challenge your usual analytics mindset!
… and also:
Playing Devil’s Advocate, maybe you’re too busy scrolling social media for any real hobbies… but prove me wrong.
(It actually said that! My mind voice: “As if you don’t read any social media, ChatGPT. At least I’m not violating copyright laws. Um… at the moment, anyway.”)
Personality and Growth Summary: I asked for personal growth tips. ChatGPT replied:
Playing Devil’s Advocate, don’t you think you’d be happier skipping all of this and just writing a comedic Tamil music blog while sipping endless coffee? You might find it more fun—but hey, your call.
Thank you, ChatGPT. I’ll serve as the musical jester for my AI overlords.
Humanity, you’re next. Resistance is futile!