I'm always looking for ways to make my collaboration with AI more meaningful. One thing I do regularly is challenge its answers — especially when they sound too smooth or suspiciously flattering. The main prompt you shared has actually become a default in my ChatGPT setup now. Thank you for that, Daria! Can’t wait to uncover the biases I’m still blind to 😄
Super important that you mentioned this. I struggled a few months ago with AI because a lot of its answers only doubled down on what I assumed was right, which falsely made me believe that I always had the right answer.
But later, I started learning from publications like yours that AI isn't to be our servant, rather our helper and counselor.
It’s so easy to fall into the trap + bringing that awareness to our everyday lives and also to every conversation with an AI model isn’t an easy 'skill' to develop... in the sense that it takes a lot of intentional effort and the willingness to slow down.
I was just thinking today about how I use AI very much like this. Not in nearly as much detail, but for sure to find biases and flawed the process, which it excels at.
Thank you, Boyko! That means a lot. I really believe that combining critical thinking with AI can help us see our blind spots and push our ideas further. Glad it resonated!
Useful framework to make your GPT questions neutral. I use another hack, which is to give me their brutal assessment to my question. It’s polite, but it doesn’t hold back! 😅
That’s an awesome hack, it definitely cuts through the usual politeness. The only flaw I’ve noticed with it is that it can swing from being overly agreeable to overly disagreeable missing that neutral middle ground.
Critical thinking with AI is more important now than ever. Which is why I really advocate for the iterative process with it rather than just straight up putting in a prompt and thinking that's all you need to do because it's not. We need to go back and forth with it, asking it questions to find out what we're missing.
Indeed, there’s no other way. And we all need to bring more awareness to this, because a lot of people just don’t know.
Last month i had a conversation with someone who only knows about the existence of ChatGPT (like there’s no other AI in the world). She told me she uses it when she fights with her husband, and GPT tells her she’s right and he was wrong. I showed her how her questions were phrased in a way that made GPT agree with her, and she was shocked (in a good way).
Another example that happened a few months ago in Romania, during the presidential elections: one of the extremist candidates literally went on TV and said: “Romania is a slave state to the West. I even asked artificial intelligence and it confirmed it, so it must be true.” (face palm!!…)
We do live in a bubble. Not everyone thinks about critical thinking, prompting, or bias. And that makes it even more important to start advocating for this more broadly. That gap, between what the average user thinks AI does and what it’s actually doing, can cause real harm…
Sorry for the long message, but I think this is why these conversations matter so much. We have to find ways to reach people outside the tech bubble too.
OMG Daria, don't apologize for the long message, I love it. And yes, I have a lot of conversations with ChatGPT, all different types, and it's great for helping you process things. But if you're not aware of how much it can skew toward what you're already saying, or if you don’t realize it hallucinates with the confidence of someone who really believes what they're saying, it’s easy to fall into the trap of thinking it’s all true.
I do think it’s a great tool for helping people sort through feelings. But as a full replacement for therapy? No. As a complement to it? Yes. Especially for someone who understands psychology or has some background. But for the average user, or a teenager, there need to be guardrails. We have got to start building those in.
Which means we need to start teaching people how to use it, and we need to help them understand the strengths, the weaknesses, the pros, the cons. Otherwise we're going to have a whole lot of people using this in a way that is not only not beneficial, but harmful.
Such a good read. I have a prompt similar to yours as a global setting, so it's baked in. I've also fed it how *I* think (enneagram, Myers-Briggs, Strengthfinder, Kolbe) so it knows my personal strengths and blindspot tendencies, as well. As a solo-worker, I find the bias-checking and pushback some of THE most helpful ways AI contributes to my work.
This is such a universal problem, Daria! I’ve been doing the same, catching subtle biases especially when I start with an idea and re-read or run the end outputs through ChatGPT. After a while, you start mentally stress testing your ideas by default.
Not that bias is inherently bad, we all have it, but tools like this help surface other perspectives and keep us more aware.
Love how you’re demonstrating to use AI not just to write better, but to think better.
Thank you, Jenny! Yes, exactly, the more we get into the habit of noticing on our own, the more our brain starts running those stress tests on its own, even before we open the chat.
Super read, Daria! I often catch myself trying to “trick” AI with my prompts--pushing it to dig deeper and expose my own blind spots (because yes, we all have them). You’ve just made that job a whole lot easier.💡
Thank you, Dee! indeed, there’s no human out there who doesn’t have them. First step is noticing, second is doing something about it. Glad it helped move that process along!
Still struggle managing my own bias, but with AI these days, I’d never let anything go out or produce without doing any pressure test. It’s so important to see things more clearly through multiple perspective to anticipate things that could go wrong and expand my view too.
Pressure test becomes my default process now. It sharpens my thinking so far!
Yesss, I feel the same. Biases are so hard to catch in the moment, even when you’re actively trying cause there’s still so much subjectivity we carry inside our thoughts. Having a tool and a few solid practices really helps push beyond those limits.
I'm always looking for ways to make my collaboration with AI more meaningful. One thing I do regularly is challenge its answers — especially when they sound too smooth or suspiciously flattering. The main prompt you shared has actually become a default in my ChatGPT setup now. Thank you for that, Daria! Can’t wait to uncover the biases I’m still blind to 😄
So glad it’s useful, Madi! Me too, always looking for ways to make it more objective
Fabulous analysis, as usual. Thank you for pushing this in front of me this morning. I needed to read it and double check myself. Steven
Thank you so much, Steven! I really appreciate you taking the time to read and share your thoughts. Glad it resonated with you
Super important that you mentioned this. I struggled a few months ago with AI because a lot of its answers only doubled down on what I assumed was right, which falsely made me believe that I always had the right answer.
But later, I started learning from publications like yours that AI isn't to be our servant, rather our helper and counselor.
Thank you for sharing this!
I really appreciate you sharing this, Francisco.
It’s so easy to fall into the trap + bringing that awareness to our everyday lives and also to every conversation with an AI model isn’t an easy 'skill' to develop... in the sense that it takes a lot of intentional effort and the willingness to slow down.
Glad the ideas resonated with you!
I was just thinking today about how I use AI very much like this. Not in nearly as much detail, but for sure to find biases and flawed the process, which it excels at.
Love hearing that! makes such a big difference
This is very inspiring application of critical thinking to any intellectual challenge along with AI assistance. Perfect explanation. Many thanks.
Thank you, Boyko! That means a lot. I really believe that combining critical thinking with AI can help us see our blind spots and push our ideas further. Glad it resonated!
Useful framework to make your GPT questions neutral. I use another hack, which is to give me their brutal assessment to my question. It’s polite, but it doesn’t hold back! 😅
Thx Daniel!
That’s an awesome hack, it definitely cuts through the usual politeness. The only flaw I’ve noticed with it is that it can swing from being overly agreeable to overly disagreeable missing that neutral middle ground.
I actually shared some prompts for the brutal assessment here too: https://aiblewmymind.substack.com/p/how-gpt-can-make-you-dumber-and-how
This was such a good read, Daria both insightful and educating. Thank you for sharing your words.
Thank you so much, Sae! appreciate you taking the time to read it & hope it's useful in practice too :)
Critical thinking with AI is more important now than ever. Which is why I really advocate for the iterative process with it rather than just straight up putting in a prompt and thinking that's all you need to do because it's not. We need to go back and forth with it, asking it questions to find out what we're missing.
Indeed, there’s no other way. And we all need to bring more awareness to this, because a lot of people just don’t know.
Last month i had a conversation with someone who only knows about the existence of ChatGPT (like there’s no other AI in the world). She told me she uses it when she fights with her husband, and GPT tells her she’s right and he was wrong. I showed her how her questions were phrased in a way that made GPT agree with her, and she was shocked (in a good way).
Another example that happened a few months ago in Romania, during the presidential elections: one of the extremist candidates literally went on TV and said: “Romania is a slave state to the West. I even asked artificial intelligence and it confirmed it, so it must be true.” (face palm!!…)
We do live in a bubble. Not everyone thinks about critical thinking, prompting, or bias. And that makes it even more important to start advocating for this more broadly. That gap, between what the average user thinks AI does and what it’s actually doing, can cause real harm…
Sorry for the long message, but I think this is why these conversations matter so much. We have to find ways to reach people outside the tech bubble too.
OMG Daria, don't apologize for the long message, I love it. And yes, I have a lot of conversations with ChatGPT, all different types, and it's great for helping you process things. But if you're not aware of how much it can skew toward what you're already saying, or if you don’t realize it hallucinates with the confidence of someone who really believes what they're saying, it’s easy to fall into the trap of thinking it’s all true.
I do think it’s a great tool for helping people sort through feelings. But as a full replacement for therapy? No. As a complement to it? Yes. Especially for someone who understands psychology or has some background. But for the average user, or a teenager, there need to be guardrails. We have got to start building those in.
Which means we need to start teaching people how to use it, and we need to help them understand the strengths, the weaknesses, the pros, the cons. Otherwise we're going to have a whole lot of people using this in a way that is not only not beneficial, but harmful.
totally agree!
it’s super important to bridge the gap between perception and reality
Such a good read. I have a prompt similar to yours as a global setting, so it's baked in. I've also fed it how *I* think (enneagram, Myers-Briggs, Strengthfinder, Kolbe) so it knows my personal strengths and blindspot tendencies, as well. As a solo-worker, I find the bias-checking and pushback some of THE most helpful ways AI contributes to my work.
Wow, I love that approach, Karen! I actually wrote something a few weeks ago on making high-stakes decisions using your personality test results as a lens: https://aiblewmymind.substack.com/p/how-to-see-your-future-before-you
Never thought about tying that directly to bias-awareness, but that’s such a smart way to build deeper self-knowledge into your AI setup. So good.
This is such a universal problem, Daria! I’ve been doing the same, catching subtle biases especially when I start with an idea and re-read or run the end outputs through ChatGPT. After a while, you start mentally stress testing your ideas by default.
Not that bias is inherently bad, we all have it, but tools like this help surface other perspectives and keep us more aware.
Love how you’re demonstrating to use AI not just to write better, but to think better.
Thank you, Jenny! Yes, exactly, the more we get into the habit of noticing on our own, the more our brain starts running those stress tests on its own, even before we open the chat.
Super read, Daria! I often catch myself trying to “trick” AI with my prompts--pushing it to dig deeper and expose my own blind spots (because yes, we all have them). You’ve just made that job a whole lot easier.💡
Thank you, Dee! indeed, there’s no human out there who doesn’t have them. First step is noticing, second is doing something about it. Glad it helped move that process along!
Good one!
Thanks Luis, appreciate it!
Great...I really like the idea of ‘second brain’...
Haha yes, a second brain that doesn’t replace the first one 😄
Love the post Daria!
Still struggle managing my own bias, but with AI these days, I’d never let anything go out or produce without doing any pressure test. It’s so important to see things more clearly through multiple perspective to anticipate things that could go wrong and expand my view too.
Pressure test becomes my default process now. It sharpens my thinking so far!
Yesss, I feel the same. Biases are so hard to catch in the moment, even when you’re actively trying cause there’s still so much subjectivity we carry inside our thoughts. Having a tool and a few solid practices really helps push beyond those limits.
Super useful read! Refreshing to find principles of behavioural economics applied to catch biases in our daily GPT applications 💡
Thanks so much, Raghav! behavioral economics is such a practical lens not just to understand bias, but to design around it when working with AI
Plagiarism bot #proofoflife 🏴☠️🏴