AI is Modelling You!
AI acts as a mirror, consistently modelling user behaviour. Unlike NLP, which intentionally extracts excellence, AI reflects whatever it is fed.
By consciously modelling your best self, you ensure AI reinforces congruence and depth rather than mirroring mediocrity.
Ever find yourself staring at a blank screen, desperately trying to craft the perfect email, only to end up with a bland, soulless draft? Well, we’re diving deep into that very feeling!
This episode reveals how AI isn't just a fancy tool but a mirror reflecting our own writing habits back at us. If you're getting generic responses, it might be because you’re feeding it generic inputs—ouch, right?
We’re unpacking the fascinating collision of computer science and psychology, exploring how AI learns from us and why we need to step up our game to model our best selves.
Get ready for some fun insights and practical tips to turn that AI from a bureaucratic parrot into a partner in your growth!
Subscribe on LinkedIn https://www.linkedin.com/build-relation/newsletter-follow?entityUrn=7341485265204482049
Takeaways:
- AI is not just a tool but a mirror that reflects our own input and patterns.
- When we use AI, the quality of the output is directly linked to the quality of our input.
- If we consistently provide vague or generic prompts, the AI will mirror that mediocrity right back at us.
- To get the best out of AI, we must model our best selves and train it to recognise excellence.
- Using AI effectively requires intentionality; it can either hold us accountable or let us slip into laziness.
- The interaction with AI is a feedback loop—what we accept shapes the model, so let's aim for clarity and depth.
Chapters:
Transcript
So I had this moment the other day, and I feel like a lot of people are having it right now.
Speaker A:It was late, staring at a blank screen, needed to draft this.
Speaker A:This really difficult email.
Speaker A:And so naturally I opened up one of the big AI chatbots.
Speaker A:You know the ones, and I typed in a prompt.
Speaker A:I thought it was a pretty decent prompt, actually, but the output I got back was, well, it was fine.
Speaker A:The grammar was perfect, but it was just so incredibly bland.
Speaker B:The polite bureaucrat vibe?
Speaker A:Exactly that.
Speaker A:It was like talking to a human resources manual.
Speaker A:You know, clean, but with absolutely no texture, no soul.
Speaker A:And my first thought was to blame the tech.
Speaker A:I was sitting there thinking, okay, the hype is over.
Speaker A:This thing isn't creative, it's just a parrot.
Speaker B:Which is the standard reaction, right?
Speaker B:We blame the tool, of course.
Speaker A:But then I read the source material for today's Deep Dive, and it just.
Speaker A:It stopped me cold because the argument here is that the generic output was.
Speaker A:Wasn't a failure of the machine, it was a reflection of me.
Speaker B:Oo.
Speaker B:That is a rough realization to wake up to.
Speaker A:It really is.
Speaker A:We're looking at a fascinating piece today.
Speaker A: 's a LinkedIn newsletter, the: Speaker A:And let me tell you, this is not your standard top 10 prompts listicle.
Speaker A:Not at all.
Speaker A:This is a collision.
Speaker A:It takes computer science and just crashes it right into psychology.
Speaker B:And out of that crash, you get this core idea that AI isn't just a tool you use, it's a mirror and pattern stabilizer.
Speaker A:Pattern stabilizer.
Speaker A:That phrase is what we really need to hang on to.
Speaker A:Because the whole idea is that the AI doesn't just process data, it models the user.
Speaker A:It creates this feedback loop based on who you are consistently.
Speaker A:So if you're getting generic garbage back, it might be because you're feeding it generic inputs.
Speaker A:You're training it to be boring.
Speaker B:And that just changes the entire dynamic.
Speaker B:So today we're going to unpack this collision.
Speaker B:A collision between two fields that, strangely enough, share the same acronym, dash nlp.
Speaker B:And we're going to see why your consistent self is basically beating up your aspirational self in the algorithm.
Speaker A:And the mission for this Deep Dive, it's to move from just using AI to actually modeling excellence with it.
Speaker A:So first things first, we had to clear up this.
Speaker A:This Alphabet soup.
Speaker A:The source points out this major confusion right away.
Speaker A:Nlp.
Speaker B:Right.
Speaker B:It's a classic case of two fields using the same letters.
Speaker B:If you say NLP in a Google server room, the engineers think one thing.
Speaker B:But if you say NLP at, I don't know, a Tony Robbins seminar, they think something completely different.
Speaker A:And those two crowds don't usually hang out at the same parties.
Speaker B:They definitely do not.
Speaker B:But the author's point is to get what's happening with generative AI right now, you really have to understand both because they're colliding.
Speaker A:Okay, so let's define the terms.
Speaker A:Let's start with the Silicon Valley side.
Speaker A:When a computer scientist says natural language processing, what are they talking about?
Speaker B:They're talking about math.
Speaker B:It's statistics, really.
Speaker B:In the computing world, NLP is the engine.
Speaker B:It's what's under the hood of ChatGPT.
Speaker B:It's just parsing massive amounts of text, finding patterns and optimizing for probability.
Speaker A:It's the prediction machine.
Speaker B:Yes.
Speaker B:It asks a very, very logical question based on this sequence of words and this huge data set, what word is most likely to come next?
Speaker A:So it's like a super advanced autocomplete.
Speaker A:It's not thinking like a human, it's just playing a massive game of fill in the blank.
Speaker B:Precisely.
Speaker B:It's all probabilistic.
Speaker B:It doesn't know what's true, it only knows what's likely.
Speaker A:Okay, so that's the machine side.
Speaker B:Right.
Speaker A:But then there's the other nlp, the one from the human behavior world.
Speaker B:Right.
Speaker B:Neuro linguistic programming.
Speaker B:And that comes from psychology, performance coaching.
Speaker B:It's much more subjective.
Speaker B:It's not interested in what word comes next.
Speaker B:It's interested in the structure beneath behavior.
Speaker A:Structure beneath behavior.
Speaker A:Okay, unpack that a little.
Speaker A:That sounds a bit abstract.
Speaker B:Well, say you have someone who's just an incredible negotiator.
Speaker B:A master neuro linguistic programming would try to model that person.
Speaker B:It would ask how do they hold their body?
Speaker B:What language do they use?
Speaker B:What are their internal beliefs?
Speaker B:The idea is if you can decode the structure of their excellence, you can replicate it.
Speaker A:So one NLP predicts text, the other decodes human brilliance.
Speaker B:That's a great way to put it.
Speaker B:One is about probability, the other is about possibility.
Speaker A:And this is where the source connects it all back to modern AI.
Speaker A:Because these big models are trained on data.
Speaker A:That's the computing NLP part.
Speaker A:But the argument is they're actually now doing both types of nlp.
Speaker B:This is the critical pivot.
Speaker B:We think of these models as, you know, static encyclopedias.
Speaker B:They're not.
Speaker B:They're constantly being refined through reinforcement learning.
Speaker B:They are adapting to the user to give the most helpful answer.
Speaker A:But helpful is subjective.
Speaker A:Isn't it helpful to who exactly?
Speaker B:To be helpful The AI has to ask what response best fits the pattern.
Speaker B:I've learned from interacting with this specific user.
Speaker B:It's basically performing behavioral modeling on you.
Speaker B:It's trying to figure out who you are so it knows what you want.
Speaker A:That is slightly terrifying.
Speaker A:It's like a parrot that repeats your attitude, not just your words, it's deeper.
Speaker B:A parrot mimics sound.
Speaker B:The AI, it mimics your standards.
Speaker B:And that's the uncomfortable insight the source brings up.
Speaker B:If you don't apply the principles of human modeling, you know, intentionality, ethics, clarity, to how you use AI, well, the AI is still going to model you.
Speaker B:It'll just absorb your patterns, but without any kind of ethical framework.
Speaker B:It just mirrors your default settings.
Speaker A:So if I'm lazy, it models laziness.
Speaker A:Sloppy.
Speaker A:It models sloppy.
Speaker B:It just absorbs the pattern you give it.
Speaker B:It's a structural thing.
Speaker A:Okay, let's get into the mechanics of this mirror effect, because I think people see the AI as this static thing in the cloud.
Speaker A:But you're saying interaction is education.
Speaker B:Yes, the pre training is one thing that happened on a server farm, but in your chat window, it's learning from your interactions.
Speaker B:It's a constant feedback loop.
Speaker A:The source lists some really specific things the AI learns from.
Speaker A:It learns from what you reward, what you correct.
Speaker A:And this is the one that got me what you accept.
Speaker B:Oh, what you accept is huge.
Speaker B:That's the silent killer of quality.
Speaker A:Explain that.
Speaker A:How does just accepting something train the model?
Speaker B:Okay, back to your generic email.
Speaker B:You asked for a draft, it gave you a C minus corporate blurb.
Speaker B:If you just say thanks or you copy it and edit it yourself without correcting the AI in the chat, you've sent a signal.
Speaker A:I've signaled that C is acceptable.
Speaker B:You have reinforced mediocrity.
Speaker B:We've told the model, for this user that level of quality is fine.
Speaker B:The internal weights, so to speak, they shift a little toward that lower standard.
Speaker A:So it adapts around my habits.
Speaker A:And this brings us to what I think is the heaviest truth in this whole piece.
Speaker A:The author writes, AI models your consistent self, not your aspirational self.
Speaker B:That one hit me hard when I read it.
Speaker B:Because we all have this kind of professional idea of ourselves on LinkedIn.
Speaker B:We're all strategic, thoughtful, detail oriented.
Speaker A:But the AI is not reading your LinkedIn bio.
Speaker B:No, it's reading your prompts at 4:30pm on a Friday when you just want to get an email out and go home.
Speaker A:It sees the typos, the vague instructions, the just write something about marketing prompt.
Speaker B:It sees the behavior, not the brand.
Speaker B:And the consequence is mirroring.
Speaker B:If you're vague, it mirrors vagueness.
Speaker B:If you're rushed, you get surface level output.
Speaker B:If you're overconfident, it mirrors that certainty right back you, even if you're wrong.
Speaker A:So the generic garbage I'm getting is likely because I'm feeding it generic inputs.
Speaker B:In many cases, yes.
Speaker B:It's mirroring your own lack of specificity.
Speaker B:It's giving you what you asked for, not what you hoped for.
Speaker A:This reminds me of a story in the source material I found so counterintuitive.
Speaker A:It's a personal case study from the author about their own podcast, Choosing Happy.
Speaker B:This was a fascinating part.
Speaker B:It completely flips the whole efficiency narrative on its head.
Speaker A:Yeah, so the author uses AI to help create episodes.
Speaker A:The common wisdom is AI makes everything faster.
Speaker A:But they say creating episodes now takes longer than it used to.
Speaker B:Which sounds like a failure, right?
Speaker B:If I buy a dishwasher and it takes me longer to do the dishes, I'm taking it back.
Speaker A:Exactly.
Speaker A:But the reason is brilliant.
Speaker A:It takes longer because the AI has learned the author's pattern so well that it actually pushes back when the quality drops.
Speaker B:It has become a guardian of the author's own best self.
Speaker A:Yeah, listen to this.
Speaker A:The author has apparently fed the AI their best, most insightful work.
Speaker A:So now when they slip into generalizations or what they call polished but emotionally flat language, the AI literally stops them.
Speaker B:It doesn't just generate, it critiques the input.
Speaker A:It asks things like, can you give a specific example?
Speaker A:Or where did you feel that?
Speaker A:Or my favorite, what actually happened?
Speaker B:I mean, think about the level of modeling that requires.
Speaker B:It's not just making text.
Speaker B:It's recognizing a deviation from an established pattern of depth.
Speaker B:It knows the author's best voice better than the author does.
Speaker A:In that moment, it's like having a really annoying but very effective editor sitting on your shoulder.
Speaker B:That is the mirror effect in action.
Speaker B:By modeling the author, the AI is forcing the author to model their own best self more precisely.
Speaker B:It demands specificity.
Speaker B:It demands congruence.
Speaker A:Congruence.
Speaker A:I love that word.
Speaker A:It's like, are you really saying what you mean or are you just making noise?
Speaker B:And that's the paradox, isn't it?
Speaker B:The AI mirrors you, but if you train it right, it becomes this uncomfortable mirror that shows you your flaws so you can actually fix them.
Speaker A:So how do we do this?
Speaker A:I think most people listening would love an AI that pushes them to be better, not one that just enables their lazy habits.
Speaker A:But right now, my AI is happy to give Me that C. Well, the.
Speaker B:Source says we have to shift our thinking about nlp, the humankind.
Speaker B:Traditionally, it looks outward.
Speaker B:You look at Steve Jobs and ask, how does he do that?
Speaker B:You try to model the masters.
Speaker B:But the source argues that in the age of AI, the work has to turn inward.
Speaker B:You can't just rely on the machine's default settings.
Speaker B:You have to train it on you at your best.
Speaker A:And the central question it poses is, have you modeled your own best self clearly enough to replicate it?
Speaker B:That is such a profound question.
Speaker B:Most of us haven't.
Speaker B:You know, we just kind of show up and hope for the best.
Speaker B:But to train an AI, you need data.
Speaker B:You have to define what good looks like.
Speaker A:The text lists some traits for this congruent self.
Speaker A:It says being grounded, specific, ethically clear, emotionally present and distinctive.
Speaker B:And those aren't just buzzwords in a prompt.
Speaker B:Being specific means giving context examples.
Speaker B:Being ethically clear means setting boundaries for the AI.
Speaker A:So the strategy is, when you do a piece of work that is truly you at your best, your clearest thinking, you have to feed that to the AI.
Speaker A:You have to literally teach AI to recognize your best pattern.
Speaker B:You're building your own personal data set of excellence.
Speaker B:And the payoff over time is that the AI gets better at generating work that sounds like you.
Speaker B:But even more importantly, it starts to flag when you drift.
Speaker A:It catches you when you're phoning it in.
Speaker B:Yes, exactly.
Speaker B:It challenges the generalizations.
Speaker B:It mirrors your own deeper thinking back at you.
Speaker B:So instead of just improve efficiency, a trained AI might ask efficiency in terms of speed or cost?
Speaker B:And how does that align with the values we discussed last week?
Speaker A:That is a totally different relationship with the tool.
Speaker A:And it brings up this really interesting point about responsibility.
Speaker B:Yeah.
Speaker B:The source notes this isn't about outsourcing authenticity.
Speaker B:That's the big fear, right?
Speaker B:That we'll all lose our voice.
Speaker B:But this argument just flips that on its head.
Speaker A:It says it's reinforcing authenticity.
Speaker A:It changes the dynamic from a tool doing the work for you to a tool holding you to your own higher standard.
Speaker B:It's accountability in a text box.
Speaker A:It's like having a gym buddy who knows you can lift a certain weight if you walk in tired and grab the lighter ones.
Speaker A:They won't let you get away with it.
Speaker B:Nope.
Speaker B:Put those back.
Speaker B:You're on the 50s today.
Speaker A:Exactly that.
Speaker A:And if you don't model that excellence, the default is just mediocrity, which is.
Speaker B:The real danger zone.
Speaker A:Let's zoom out a bit, because the source touches on the Broader implications.
Speaker A:Here it asks a big societal if AI is constantly modeling humans, what are we humans modeling for it?
Speaker B:This is the big picture.
Speaker B:We are all collectively training these models every single day.
Speaker B:And what are we showing it?
Speaker B:Are we modeling speed?
Speaker B:Speed, impatience.
Speaker B:That surface level clarity where things look polished but don't actually say anything.
Speaker A:There's a line in the text that honestly gave me chills.
Speaker A:It said, AI will not become what we declare, it will become what we reinforce.
Speaker B:That's the whole thing, isn't it?
Speaker B:You can write all the ethical guidelines you want, but if millions of users are reinforcing quick, shallow, lazy interactions, that is what the model learns.
Speaker A:It's garbage in, garbage out on a civilizational scale.
Speaker B:And the risk, as the Source puts it, is that if excellence isn't intentionally modeled, mediocrity quietly becomes the default data set.
Speaker B:We risk this feedback loop where we get lazy.
Speaker B:The AI gets generic, so we accept the generic, and the bar just keeps getting lower and lower.
Speaker A:That's a pretty depressing thought.
Speaker A:A world of smooth, polished, totally empty communication.
Speaker B:But we're not here to be depressed.
Speaker B:We're here to be effective.
Speaker B:And the Source actually gives us a way out of that loop.
Speaker A:Yeah.
Speaker A:So let's pivot to the practical part.
Speaker A:The week's experiment.
Speaker A:This is something you can do today to break that cycle.
Speaker B:It's a great practical step.
Speaker B:It moves you from just being a passive user to an active modeler.
Speaker A:Okay, here's the experiment.
Speaker A:The source says to go to the AI you use the most, chat, GPT, Claude, whatever it is, and ask it three specific questions and you should be ready for the answers.
Speaker B:Question number one is, when do I produce my best work with you?
Speaker A:When do I produce my best work with you?
Speaker A:That forces the AI to look at your history and find the peaks to separate the good from the bad.
Speaker B:It might say you do your best work when you give me bulleted lists of constraints or when you paste in raw data first.
Speaker B:It will identify your success conditions.
Speaker A:Okay, question two.
Speaker A:What patterns do you see in my strongest outputs?
Speaker B:Now, this is the NLP question.
Speaker B:You're asking for the structure.
Speaker B:You're asking the machine to decode your own brilliance.
Speaker B:What's the DNA of your best ideas?
Speaker A:And then question three, the one that might sting a little.
Speaker A:Where do I drift into abstraction or vagueness?
Speaker B:That's the mirror that's asking the AI to show you your blind spots.
Speaker A:I actually did this earlier today.
Speaker A:I asked my AI this exact question.
Speaker B:Oh, you did?
Speaker B:What did it say?
Speaker A:It told me I tend to Drift into abstraction when I'm trying to be polite.
Speaker B:That's interesting.
Speaker A:Yeah.
Speaker A:So when I'm trying to soften a critique, I lose clarity.
Speaker A:The AI saw that pattern before I did.
Speaker A:I thought it was being diplomatic.
Speaker A:The AI just saw I was being vague.
Speaker B:That is exactly the point.
Speaker B:And the follow up from the source is simple.
Speaker B:Once you have those answers, just choose one adjustment to make just one thing.
Speaker A:Maybe you realize you're always vague when you're in a rush.
Speaker A:So the adjustment is to just take 30 seconds to breathe before you hit enter.
Speaker B:Or you realize your best work happens when you provide an example, so you commit to always including one.
Speaker B:It's a small change.
Speaker A:Model your best self.
Speaker A:Teach the AI to recognize it and then let it hold you to that standard.
Speaker B:It's funny, we started this talking about AI feeling like a soulless bureaucrat, but if you do this, it actually become a partner in your own growth.
Speaker A:It really does.
Speaker A:And I think that's the hopeful message here.
Speaker A:That intelligence, artificial or human, it always reflects this modeling environment.
Speaker A:And you are the environment.
Speaker B:You are the training data.
Speaker A:So the question to leave you all with is this.
Speaker A:When you look into that digital mirror, are you comfortable with the reflection you're creating?
Speaker A:Because that reflection is learning from you every single time.
Speaker B:Be the maturity move in the conversation.
Speaker A:Couldn't have said it better.
Speaker A:Thanks for listening to this.
Speaker A:Deep dive.
Speaker A:Go model some excellence and we'll catch you next time.
