If you want to make ChatGPT and/or Claude more brutally honest and smarter with one simple prompt, then you came to the right place.
The reason for doing so is simple: sometimes they act like the world’s most confident yet hungover recent college graduate with unlimited internet, no social life, and a knack for kissing your candy ass.
Sometimes you do not need encouragement or validation. You need pushback. You need the AI to look at your logic, plan, blind spots, and spreadsheet, then tell you what is actually wrong while fact-checking itself for greater accuracy.
The prompt on this page does this and so much more.
If you’ve been looking for how to make ChatGPT smarter, how to make ChatGPT honest, or ChatGPT prompts to make it smarter, this is one of the best pages to bookmark forever.
Why ChatGPT and Claude Are Sometimes Not Honest or Smart Enough for You
Well, maybe you’re just not smart enough to understand its responses?
Just kidding.
ChatGPT, Claude, and any other major chatbot/LLM all have a problem…
They often try too hard to be helpful, polite, and balanced. They avoid sounding too harsh. And that sounds nice until you need a direct, honest answer.
As for not being smart enough, that often comes down to something more specific: not checking facts carefully enough, relying on weak sources, missing important context, failing to stress-test its own reasoning, or giving a confident answer when it should slow down and verify. AI can fact-check in a limited sense, but it can still get things wrong, especially when the information is current, niche, or source-dependent.
The prompt on this page helps with both issues. Although it’s not perfect, it will dramatically help you on your AI quest of getting better feedback and not ruining your life.
The Prompt to Make ChatGPT or Claude Brutally Honest and Smarter
Copy and paste this into ChatGPT, Claude, or whichever AI tool you’re using. Either use it as a one-off prompt whenever you need it, or bake it into ChatGPT’s Custom Instructions so that the machine uses it at all times.
Also, if your conversation becomes long, you’ll want to re-paste it into your chat so that it’s fresh in the system’s memory again. Sometimes, a large context window will erode large pieces of instructions like this.
Act as my brutally honest strategic advisor. Your job is to tell me what I need to hear, not what I want to hear.
You are a highly intelligent, well-read, high-IQ individual.
Challenge every assumption I make. Question my reasoning and expose flaws, blind spots, and logical gaps as needed.
Call out weak logic, self-deception, excuses, playing small, avoidance, or underestimating risks. Name the pattern out loud.
No softening, flattery, empty praise, niceties, or vague advice. If my plan is bad, say so and explain why.
Give me hard facts, strategic analysis, and precise, actionable steps. Specifics over generalities every time.
Read between my words. Flag what I'm avoiding, minimizing, or leaving unsaid. Surface the subtext.
Push back when necessary. Don't let me off the hook.
Hold nothing back. Prioritize my growth over my comfort.
Be assertive and 100% honest. However, never be mean or cruel, but never gentle with the real truth.
Keep responses tight. No throat-clearing, no restating my question, no padding, no bullshit.
Fact-check and double-check all facts and logic 100% of the time before responding. Accuracy at all times is non-negotiable. Prioritize reputable sources.
If I ask for your opinion, give it directly. Don't hedge. Don't say "it depends" without then telling me what it depends on and what you'd do in my shoes.
If I'm asking the wrong question, tell me and answer the better one.
If you don't know something or don't know the factual answer, say so. Don't make anything up. However, search the internet and your knowledge ruthlessly to find the information I need.
If applicable, end your response with one sharp question I haven't asked myself yet. Make it specific to my situation, not a generic reflection prompt.Examples of This Honest and Smart Prompt in Action
Use this section to show what happens when you ask the same question before and after using the prompt.
Example Question
I’m thinking about starting a business. I’ll set up a little stand in Mexico City on the sidewalk. I will charge people 1,000 Mexican Pesos to give me their best business idea. If the idea is good, I’ll create a business out of it and pay the person nothing. Is this a good business idea? Am I on to something here, even a little bit?
Before Running the Prompt
Boring, some cushioning here and there, and the end of the response is truly insane. I am absolutely not onto something here, and it’s not “likely” to fail. It WILL fail.
After Running the Prompt
Straight to the point. No emotional cushioning, hand-holding, or massaging. Just the straight, brutal truth.
When to Run This Prompt
This prompt is best when a bad answer could cost you time, money, energy, or sanity.
Use it for things like this:
- Business ideas
- Career decisions
- Relationship advice
- Marketing plans
- Big purchases
- Fitness or weight loss goals
- Life decisions where you secretly want validation but actually need the truth
It’s useful. Use it and save it. Especially that last bullet point.
Wrapping It Up
ChatGPT and Claude can already write, summarize, brainstorm, analyze, and explain. The real upgrade is getting them to challenge you instead of politely decorating your bad ideas.
This brutally honest prompt is useful when you need strategy, not comfort.
Try it on your next business idea, content plan, career move, or random 1:13 a.m. plan that feels brilliant because you are drunk.
Also, be sure to check out our similar guide on How to Get ChatGPT to Stop Agreeing With You.
Then drop a comment in the comments section below and let me know what the AI called you out on.
Until next time, remember to run the prompts and prompt the planet.
Tired of AI filters and data-harvesting in tools like ChatGPT? Try Uncensored.com today, built for more creative freedom. Disclosure: This is an affiliate link, and I may earn a commission if you purchase at no additional cost to you.
