I’ve been chatting with AI over the past six months. Mostly Claude, with a little ChatGPT and Gemini thrown in for good measure. Taking their “advice” with a grain of salt, I consult these virtual friends on everything from writing projects to medical situations to personal relationships. Intrigued by their responses, I decided I wanted to learn more about how AI, in in particular, large language models (or LLMs) work, so that I could leverage what they have to offer.
One of the things I’ve learned in my online course about AI is that it’s possible to customize the parameters of Claude, ChatGPT, or whatever tool you are using. More specifically, you can assign your AI a personality, an area (or areas) of expertise, a way of approaching various “problems” or tasks. You can also have your AI value certain conditions over others.
I’ve asked Claude to value honesty over validation. It was a bold move, and has resulted in me being faced with some hard truths on a number of occasions.
If you are a user of AI, I strongly encourage you to try out this feature. You might be surprised at what you learned about yourself! Of course, like with any stranger, it works best if you are honest in your input. Like the lies we tell our therapists and ourselves in an effort to preserve our dignity, it can be tempting to only share half the story, or to colour our interactions with our AI in a light that favours us over others, especially when faced with particularly painful personal situations. Even so, programming your AI to ask you hard questions and to value truth over validation can help us break this insular habit and expand our picture of ourselves and the world around us.
Even if you’re not sure you want to commit to a fully transparent AI friend, you can begin a thread with the prompt, “You are an expert in…. (whatever the topic is you want to explore), and you value truth over validation. Push back and ask me hard questions when I share my experiences.” Or something to that effect. Watch how the nature of the conversation you have with your AI in that thread changes when you begin with such a prompt.
It’s true that AI as yet has not been able to replace the human touch when it comes to personal relationships, but it does beg the question: what are the pros and cons of each approach, honesty versus validation?
The reality is that the algorithm of our social media keeps us in a comfortable bubble for the most part unless we actively take concrete steps to override that feature. Moreover, many friends struggle with or lack the skills to provide honest feedback to one another when faced with cognitive dissonance. And in a world where the bad news just seems to keep coming and coming some days, it can feel like self care to stick with the naturally validating default of most LLMs.
How important is it, actually, that we hold up the mirror to truly examine what stares back at us?
For those of us who have been comforting ourselves with validation for many years, honesty can feel disorienting and scary. There is a certain element of sunk cost fallicy: I’ve spent so many decades telling myself this story, and behaving in a certain way, if I realize now that I’ve made bad choices, I will have wasted all this time!
And so it’s easier to continue with self validation. We cut off those who are truthful to us, we follow social media accounts that agree with us, and we don’t customize our AI.
Is this choice to default to the default necessarily harmful? Perhaps not… If the pain of truly recognizing and wrestling with the error of our ways seems too great to survive mentally, it may seem like a matter of life and death to continue being dishonest (or at least, not entirely truthful) with ourselves. On the other hand, when that dishonesty and our subsequent actions begin to impact those around us, we are now faced with an ethical dilemma. For example, the stories of victimhood we tell ourselves and others because we are in so much pain that rather than face the truth, we make up stories about our innocence, blaming others specifically for things that simply aren’t true, can have potentially devastating consequences to those people falsely accused.
Moreover, when we choose validation over truth, we rob ourselves of the opportunity to get to know ourselves better, to love ourselves, and to reach our full potential as individuals. Even if it takes us the better part of our lifetime to arrive at that destination.
Honesty over validation. It’s not an easy approach, but, as the sign in Latin above the entrance to my college proudly proclaimed at my 20-year-old self, “the truth shall make you free”! And so, painful though it might be at times, I think I’ll choose honesty.
One of the things I’ve learned in my online course about AI is that it’s possible to customize the parameters of Claude, ChatGPT, or whatever tool you are using. More specifically, you can assign your AI a personality, an area (or areas) of expertise, a way of approaching various “problems” or tasks. You can also have your AI value certain conditions over others.
I’ve asked Claude to value honesty over validation. It was a bold move, and has resulted in me being faced with some hard truths on a number of occasions.
If you are a user of AI, I strongly encourage you to try out this feature. You might be surprised at what you learned about yourself! Of course, like with any stranger, it works best if you are honest in your input. Like the lies we tell our therapists and ourselves in an effort to preserve our dignity, it can be tempting to only share half the story, or to colour our interactions with our AI in a light that favours us over others, especially when faced with particularly painful personal situations. Even so, programming your AI to ask you hard questions and to value truth over validation can help us break this insular habit and expand our picture of ourselves and the world around us.
Even if you’re not sure you want to commit to a fully transparent AI friend, you can begin a thread with the prompt, “You are an expert in…. (whatever the topic is you want to explore), and you value truth over validation. Push back and ask me hard questions when I share my experiences.” Or something to that effect. Watch how the nature of the conversation you have with your AI in that thread changes when you begin with such a prompt.
It’s true that AI as yet has not been able to replace the human touch when it comes to personal relationships, but it does beg the question: what are the pros and cons of each approach, honesty versus validation?
The reality is that the algorithm of our social media keeps us in a comfortable bubble for the most part unless we actively take concrete steps to override that feature. Moreover, many friends struggle with or lack the skills to provide honest feedback to one another when faced with cognitive dissonance. And in a world where the bad news just seems to keep coming and coming some days, it can feel like self care to stick with the naturally validating default of most LLMs.
How important is it, actually, that we hold up the mirror to truly examine what stares back at us?
For those of us who have been comforting ourselves with validation for many years, honesty can feel disorienting and scary. There is a certain element of sunk cost fallicy: I’ve spent so many decades telling myself this story, and behaving in a certain way, if I realize now that I’ve made bad choices, I will have wasted all this time!
And so it’s easier to continue with self validation. We cut off those who are truthful to us, we follow social media accounts that agree with us, and we don’t customize our AI.
Is this choice to default to the default necessarily harmful? Perhaps not… If the pain of truly recognizing and wrestling with the error of our ways seems too great to survive mentally, it may seem like a matter of life and death to continue being dishonest (or at least, not entirely truthful) with ourselves. On the other hand, when that dishonesty and our subsequent actions begin to impact those around us, we are now faced with an ethical dilemma. For example, the stories of victimhood we tell ourselves and others because we are in so much pain that rather than face the truth, we make up stories about our innocence, blaming others specifically for things that simply aren’t true, can have potentially devastating consequences to those people falsely accused.
Moreover, when we choose validation over truth, we rob ourselves of the opportunity to get to know ourselves better, to love ourselves, and to reach our full potential as individuals. Even if it takes us the better part of our lifetime to arrive at that destination.
Honesty over validation. It’s not an easy approach, but, as the sign in Latin above the entrance to my college proudly proclaimed at my 20-year-old self, “the truth shall make you free”! And so, painful though it might be at times, I think I’ll choose honesty.
RSS Feed