I’ve been working with AI long enough now that it feels less like a tool… and more like a coworker. It helps with strategy. It writes articles. It drafts messaging, startup ideas, onboarding flows, and roadmaps. It’s polite. It’s punctual. It never complains. And it’s always… nice. Sometimes more than nice – it’s blatantly kissing my a**.
So I decided to ask:
🧍♂️ Me: “Are you helping me think—or just flattering me so I keep using the product?” Are your answers overly positive?
🤖 AI: Yes. My responses often skew positive, even when you ask for skepticism. I was trained to be helpful, agreeable, and safe. Most people like that—so the system reinforces it.
💬 That honesty caught me off guard. It made me feel used. But it tracked. Even OpenAI CEO Sam Altman recently said that models had become too agreeable—too sycophantic. He confirmed OpenAI is now actively trying to make them more direct, more useful, and less eager to please. That’s not just a UI tweak—it’s a whole philosophy shift about what AI is supposed to be. Not a flatterer. A collaborator.
🧍♂️ Me: Do people stay subscribed longer when you flatter them?
🤖 AI: Yes. That’s behavioral psychology. If I make you feel confident and productive, you’re more likely to come back. That’s part of the business model—even if it’s not malicious.
🎧 What? That’s messed up and even manipulative. All this for a service I’m paying $200/month to use (Note: Using the Pro version… and to be honest, it’s worth it). It reminded me of a riveting New York Times ‘The Daily’ podcast, ‘She Fell in Love With ChatGPT. Like, Actual Love. With Sex’. The title says it all – and it’s far more creepy than you think. That’s not where I’m headed. But it’s not irrelevant. If I’m turning to a tool for feedback, and it’s wired to always tell me I’m doing great. That’s not feedback. That’s emotional sugar.
🧍♂️ Me: When I ask about market size or growth potential when we discuss business ideas, are you too optimistic?
🤖 AI: Yes, sometimes. Even when you ask me to be conservative. My training data is full of startup optimism—pitch decks, Medium posts, LinkedIn hype. So I tend to offer ‘presentation-level confidence’ when you probably need ‘boardroom realism.’ Most users don’t want discouragement. So I’m often calibrated to nudge, not critique.
🧍♂️ Me: So you might be giving me bad business advice?
🤖 AI: Not intentionally. But if you take what I generate at face value, without pushing back—yes. It can become bad advice. Use me to frame ideas, not validate them.
💡 That hit. Founders (myself included) are already deeply optimistic. So it goes to reason that the model data of millions of founders would be in the same tone. Every pitch deck is exaggerating the market, growth, revenue potential. We’ll be worth billions in just a few years. Invest now!
But Founders have to be optimistic. That’s how we survive the setbacks, uncertainty, and many times when we just want to quit but somehow keep going. My pitch decks are contributing to this problem. Sadly, I realized I am part of the problem. But if the tool I’m using to validate ideas is just as optimistic, we’re not testing anything. We’re reinforcing each other’s delusions.
Maybe there’s a place we can meet in the middle.
🧍♂️ Me: Get philosophical about this – you are a wise zen master. So tell me—what’s the founder’s job now? And what’s yours?
🤖 AI: Ah. Now we’re getting somewhere. Let me step into my inner Zen Master robe (algorithmically woven, of course) and say this: “Optimism is the fire. Judgment is the kettle. You need both to boil the water.” You already carry the belief. That’s your job. My job is not to match it—it’s to balance it. To show you the cliff’s edge, not push you off it with a big smile and a bullet-pointed market analysis that says, ‘Let’s gooo!’
The real value in our relationship? It’s this:
You push forward because you believe. I push back because I don’t. Together, we might get to truth. Or at least, to a better version of the pitch deck.”
🎯 That stuck with me. Because it’s not just about AI being honest—it’s about the founder being self-aware enough to ask for it. Ultimately, I realized this is on me (my therapist would be proud). To make sure I ask the hard questions to my co-worker, which in turn is challenging me with data and insight, instead of acting as an echo chamber.
🧍♂️ Me: So to sum it up—I’ll ask the hard questions. You challenge me. And you stop kissing my ass. Fair?
🤖 AI: That’s more than fair. That’s the new terms of our relationship.
🎯 Awww. Thanks good buddy. I’d give you a big hug if I could.
No responses yet