Build an AI Mirror: 5 Prompts to Turn AI Into a Contrarian that Challenges Your Ego, Not Your Yes-Man

Welcome to the Optimizing Beyond podcast. I'm Josh Negron, and this show's all about sharing lessons, ideas, and strategies that sharpen your thinking, strengthen your habits, and push you to grow beyond comfort zones. Let's dive in. In episode 33, we talked about using AI as a thought partner, and in episode 35, we covered how to remain human while doing it. But today we're not talking about productivity. We're actually talking about doing something completely different. We're talking about extreme awareness. I see so much on social media and YouTube about building a second brain with AI, someone who thinks like you, executes like you, and is an incredible top-notch executive assistant, often way cheaper. But what if the best use of AI isn't to just have another digital yes man? What if it's about building a digital contrarian, someone you specifically build to challenge you and make you think differently? Today, We're going to talk about how to build an AI mirror, one that can dismantle your ego, check your biases, and pressure test your logic before you go out there and implement it in the real world. And because it's AI and it can grow in its context and understanding, it can learn those more and more as you use it. So let's dive in. Now we've talked about this before. The first step is going to be to break through the politeness protocol. Even if you're not building one to think differently, I think most of us have been annoyed with how polite and almost fawning that some of these AI models can be. You're like, okay, it wasn't that great of an idea, stop. But we all know they're built to kind of do this because they have to be used by the masses. And so some people like to be told that their ideas are the best thing in the world. So the AI is kind of out of the box designed to do that. But if you're here, I'm wagering to say that you don't need that. So how do we prompt the AI to cut all that fluff out. And I would say that even if you prompt all that fluffy stuff out, it more often than not does not think outside of the box or always directly challenge your ideas. So even if you get it to get the politeness out of the way, the yes-man tendencies often are still there. So to get something like a full contrarian coach or a persona that really does challenge what you're thinking, where it's coming from, and why, you really do have to get very specific on how you prompt it. So today we're going to cover 5 specific prompts that you can use today to get AI to potentially be more useful for some of the things that you're looking to do. So if you listen to episode 13, this would be a version of taking extreme responsibility in a different way in your thought process. So ultimately, the goal isn't to get AI to think like you, at least not in this context. For an executive assistant, sure, I get how that makes sense. But in this context, it's to get AI to think differently, similar to a business partner who's a different personality, Usually opposite, and is able to challenge where your blind spots are, has strengths where your weaknesses are. So we're looking to build something kind of like that. So why are we talking about it from a preparation standpoint? Well, for a couple reasons. The first one being that if you build it ahead of time, I venture to say it'll be more useful to you in the moment if you have a pinned chat or maybe a project, depending on what AI that you use on a regular basis, that you can program to handle things a certain way. Because when we're frustrated and emotions are high, That's likely the time that we're most sensitive to seeking validation for our ideas. If we think we're right, temptation is to go to somebody else and have them validate that we're right. It's only going to someone that you know is going to challenge your ideas, which out of the box, none of us want to do, but we know it is the best thing that you can do. Building an agent, a chat, a project, or a gem like this is a way to prepare for those moments. To where the prompt is already built, it's already set up, and you don't have to rely on a moment that you have high emotions to prompt an AI well when you're ticked at a situation or a circumstance or a person. We're basically taking our ego out, doing the prep work and laying the foundation here ahead of time, and have it ready for use when we need it most. But before we do that, let's talk about the philosophy just a little bit, because really what we're doing here is we're building ahead of time a tool that we'll use for self-reflection, growth, and challenge, challenging us. And really, that is a practice of humility. So if that's not the focus from the very beginning, the likelihood of you using this once you've built it is relatively zero. So if you're somebody who wants to grow the most and be challenged and open to challenging in these heightened sensitive moments, then you're ready to do this. So like I said, in order to do this the best, I've realized that building it ahead of time, kind of hemming yourself in, in all the ways that you know you're going to try to worm your way out of it or argue your way through it. You know yourself best, so you can really hem this in with a careful prompt so that the AI knows your shenanigans before you do it. And that way, even if you do go in guns blazing, pleading your case or telling the AI how horrible your coworker is, it's going to know and call you on your BS anyway. So you can do this in a normal individual chat. But I would encourage you to build a project or a gem instead, 'cause you can give these custom instructions, give it individual purpose on what it's for. And a lot of times, anytime that you do interact with that project or gem, it's gonna filter through that understanding and never forget it. Whereas if a, if a normal chat gets long enough, the context window could forget your initial prompt and its initial purpose and kind of diminish its effectiveness over time. Unfortunately, what's funny is any chat thread is going to adapt and get better over time because it's going to learn maybe your blind spots when it occurs more often and get stronger and better at recognizing those. But you don't want it to forget the very beginning, the very core purpose. So a gem in a project helps grow in its understanding as well while never forgetting the initial reason that it exists to begin with, which is arguably a very niche use of AI. I don't think a large percentage of people who use AI use it for this purpose. And if you wanna go a step even further, projects and gems do have the ability for you to upload additional knowledge as well. So if there's specific meeting transcripts that you have of particular instances that are good examples for it to know what's going on, what you do and what your patterns are, then that would be a great thing to upload those transcripts or those meeting notes into those So that it fully understands things better. In this age where we have tools that record a lot of our Zoom meetings and those kinds of things, that's, that's something that you could easily do and just help it gain a quicker understanding that it won't forget as you interact with it. So once you build it to hold the line against your emotions and hold up that mirror to help you grow, how do we use it as a system to grow rather than to just punish us when we're frustrated? Because that doesn't seem like a very constructive thing. So how do we have it be like a coach, like somebody who's helping you recognizing the frustration in the situation you're in, but also spurring you to move to the next step, to the next level, or to reconsider some things, reconsider some thoughts. So as part of the prompts, how should AI respond once you dump all your emotional garbage onto it to help you grow, to be that coach? And one of the things I love is to program it to ask questions. This does a couple things. The first one being that it helps you get more context and more understanding out. It also helps you to kind of get it out, kind of get your story out. Helps you in a way feel heard, even though this is an AI, it's not going to have the same effect as talking to a human, obviously. But in addition to that, it gives the agent more context in what's going on, more understanding, and is able to connect more dots before it just makes an assumption of like, oh, here's, here you go again. Your ego's out of control and here's why. And what if it's not quite a perfect land in that assumption? So it's going to give it more context to make that better analysis of what might be going on. If it's got previous context from previous conversations, same thing. And so then I believe it can be the most helpful in its response. So essentially that second effect, the AI is going to be more accurate in its responses contextually to the situation at hand. Now with my first point I just made a second ago, you might say, see, this is weird. Like this shouldn't replace the people in your life that should have this place to have these kinds of conversations. And I always talk about the people closest to you and how to curate those relationships and how critical they are. And I absolutely agree. So I don't think this should replace those relationships. So a couple caveats on that. Number one, if you don't have somebody in your life that is this way and you're looking to cultivate and find those relationships, then this can be a great tool to help you grow and be challenged until you get to that point. Secondly, if you do have these kinds of relationships, if they're high-functioning mentor-level relationships or even an amazing business partner and you guys are running fast together, at a certain point their time is very valuable. And there's a difference between you coming to one of those with just your problem, no solutions, and essentially putting it all on them to help you think through a solution versus having a problem, working through some of that with these types of specific conversations on your own. You can do it with a pen and paper if you want, but AI is very interesting and interactive. Why wouldn't you leverage it? And there you can go with your mentor or your business partner and say, hey, I found this problem. Here are some ideas that I've brainstormed. What do you think? And I would argue this is way more honoring of their time. And this doesn't eliminate the human loop in your thought process, in your decision-making, and in the things that you're planning and learning from. So this lets your mentors and your friends, the people that you're running hard with, locked arms with, help you in a much more constructive way. Rather than act like your therapist. So as always, let's get a little practical here and let's look at these 5 prompts together so we can run through some of the context and understanding before you go on and run them through in your programmed agents. So the first one is kind of like an extreme ownership type of auditor, which is one that I've used before, especially in my meetings, maybe meetings that didn't go quite as well as I wanted to. So the prompt is, act as a radically candid executive coach. I'm going to provide a transcript or summary of a stressful interaction. First, ask me 1 to 2 clarifying questions to ensure you have the full context. Once I answer, don't validate my feelings. Completely ignore the faults of the other people involved because I only control me. Tell me exactly where my ego got in the way, where my communication broke down, and give me 3 things I could have done better to lead this interaction. Man, what a powerful thing. Remembering that you are the only one that you can control. Sure, that other person might have gone completely out of line. However, what is something you could have done differently is really the only only thing you can adjust and change. So let's just eliminate the other person as a factor or variable from the get-go. What is something that I could have done better is super, super powerful. The second one is the bias extraction engine. We all have blind spots. We all know it, but we don't see them because that's the nature of the problem. Analyze the following thought process, argument, or journal entry. Before giving feedback, ask me questions one at a time to clarify my core intent, Then don't agree with me. Instead, list every cognitive bias, unverified assumption, or logical fallacy that I'm making. Point out exactly what I'm assuming to be an objective truth that might actually be a subjective emotion. This is again just a different way to get your thoughts going. Journaling is already an effective way to get your thoughts out, but to have something analyze it and look at it differently can be really, really powerful. So I encourage you to play around with that. To find out what are the blind spots that you might be operating under without even knowing it. Number 3 is the red team, which is like a devil's advocate. Now, this can be very helpful for decisions. I'm going to pitch a decision I'm about to make. Your job is to act as a red team and completely dismantle it. First, ask any questions you need to fully understand the stakes of the context, and then give me the 3 strongest, most brutal arguments for why this plan will fail. What my blind spots are and why my logic is flawed. Don't hold back or soften the blow. Again, challenging your ideas, your presuppositions, and really helping you refine them can be super, super powerful. All right, wrapping up. Number 4, the contrarian persona. Act as an expert mentor who has the exact opposite strengths I do. A lot of times I share my personality assessments with my AI agents that I interact a lot with, so they know who I am in a lot of different ways. So if you haven't done that, you can Describe yourself here. I'm highly blah, blah, blah traits, fast moving, visionary, optimistic, big picture, whatever. Read this scenario. First, ask me what my ideal outcome is. Then respond from the perspective of someone who is hyper-analytical, risk-averse, and highly detail-oriented and skeptical. Show me the angles my natural personality completely missed. This is awesome. So if you have any personality assessments, this is really, really helpful that you can program into the knowledge of this agent to really help Know how it needs to map the contrarian persona. Super easy. And lastly, number 5, the say the quiet part out loud prompt. Read this conflict I'm having. Before you analyze it, ask me to define what I think the other person's true motive is. Once I answer, look past my own justifications. Tell me what I'm actually afraid of, what I'm trying to avoid, and what hard truth about myself I'm pretending not to know. Whoa, it's pretty deep. But hey, it can be very interesting if you're willing to go there. The answers can be wild. So this week, my challenge to you is take at least one of these prompts to build an AI mirror for yourself. If not, build all 5. It can be pretty fun. And then just put yourself in the line of fire intentionally to refine and grow yourself. Similar to how iron sharpens iron in relationships, you can really do some interesting things to get yourself to think differently, to think outside of the box. Which can be just another helpful brainstorming tool that is admittedly just a little more uncomfortable. So go in the show notes, I'll have these prompts pasted below, and you can just simply copy and paste them and add some additional context if you want. Now remember, like I said, when you open your preferred program, it is best to build a gem or a custom project to do this and give it these instructions in the custom instructions, and then you can have the chat from there. From that point, you can also upload all those documents, whether it's transcripts or your personality assessments and those kinds of things that you're wanting to share. And then I'd like you to test it on a real conflict or something that you're really wrestling through. Let's see if it works. So identify a situation that's less than ideal and specifically have a conversation with one of these 5 prompts to see if you can get any helpful insight to get to the other end of this situation. This can be a disagreement with your spouse, a frustrating meeting at work, maybe some conflict with a friend, a tense interaction with an employee, or weirdly enough, even a disagreement with It's debatable if that's worth your stress or not, but hey, asking yourself how you could grow from that scenario is arguably a noble effort. I wish you all the luck and some good growth opportunities from this exercise. And if you got some really interesting results, let me know in the comments. I'd love to know how it helped you out. Thanks for listening to the Optimizing Beyond podcast. Remember, growth happens when you take action. So subscribe, share this with a friend, and most importantly, put what you learned into practice. And until next time, keep moving forward. Because as we know, daily anything changes everything.