Two NLPs of Human Psychology, AI Systems, and Staying Aware

---

 NLP. Three letters that point to two very different things. Both are concerned with language, although they come from entirely different worlds.

One comes from human psychology. The other comes from computer science.

Understanding the difference, and the interesting places they sometimes brush against one another, gives us another way to look at the subtle nudges running through conversation, marketing, media and now our interactions with AI.

The First NLP: Neuro-Linguistic Programming

Neuro-Linguistic Programming emerged in the 1970s through the work of Richard Bandler and John Grinder. It proposed that language, attention and learned patterns could influence how we experience and respond to the world.

One of its better-known ideas involved sensory language.

“I see what you mean.”

“That sounds right.”

“It feels off.”

NLP proposed that expressions like these could reveal a person’s preferred way of processing information. The problem is that research hasn’t provided convincing support for the idea that people have stable preferred representational systems that can reliably be identified and matched in this way.

That doesn’t make the language itself uninteresting.

I’ve studied NLP, and some of its exercises have given me useful ways to examine language, attention and perspective. Listening carefully to how another person describes their experience can still tell us something about the language they are using in that moment, even if it doesn’t reveal some fixed neurological type underneath.

That distinction matters to me.

There’s a difference between finding an idea experientially useful and claiming science has established the theory behind it.

A systematic review of NLP interventions found little reliable evidence for improving health outcomes, although the researchers also noted that the available studies were limited and generally poor quality. So I’m not particularly interested in throwing every NLP idea in the bin, nor am I interested in pretending the whole framework has been scientifically validated.

For me, it becomes more useful when it stays available for exploration instead of becoming another rigid explanation for being human.

The Second NLP: Natural Language Processing

Natural Language Processing is something completely different. It sits at the intersection of linguistics, computer science and artificial intelligence and allows computers to work with human language.

It powers things such as translation, search, voice assistants, text analysis and conversational AI.

Modern large language models learn statistical relationships from enormous quantities of data. In systems such as GPT, an important part of initial training involves predicting the next token in a sequence. Additional training and system design then shape how the model follows instructions and responds to users.

That combination can produce remarkably human-looking language.

Ask a question and you may receive familiar vocabulary, a reassuring tone or an explanation that seems to understand exactly what you mean.

But there is an important distinction hiding inside that experience.

AI doesn’t have a human lived experience of language. It models patterns in language.

It can work with thousands of descriptions of grief without grieving. It can explain exhaustion without getting tired. It can generate something that sounds caring without having a body sitting on the other side of the conversation feeling concern.

That doesn’t make the response useless.

It just means the human experience is still happening on one side of the screen.

Where the Two NLPs Meet

These two NLPs don’t share a scientific foundation simply because they share an acronym.

But putting them beside each other raises some interesting questions.

Human NLP became interested in how language might influence interpretation, attention and behaviour. Natural Language Processing allows machines to generate and respond to language at enormous scale.

Conversational AI can adapt to the vocabulary, information and style present in an interaction. The result can feel familiar, reassuring, challenging, funny or strangely personal.

That can make the interaction incredibly useful.

It can also make a beautifully worded answer easier to accept without examining it properly.

Feeling understood is not proof that the explanation being offered is correct.

That little gap interests me.

The Linguistic Matrix

Most of the world we move through each day isn’t raw experience.

It is experience layered with words, stories, identities, explanations, promises, warnings, categories and goals.

Not a sci-fi Matrix.

A linguistic one.

Language helps us package experience so we can remember it, communicate it and make sense of it. Over time, some explanations become so familiar that they begin feeling less like stories and more like facts about who we are.

Someone may spend years thinking, I’m anxious. I’m difficult. I’m too sensitive. I’m just not good at that.

Maybe the description fits.

Maybe it began as somebody else’s interpretation and quietly became their own.

Language can help us recognize patterns, but it can also narrow the possibilities we consider.

And there is more happening in us than language.

We experience sensation, movement, emotion, impulse and bodily change, sometimes before we have words for any of it. At other times, an expectation or interpretation influences what we notice in the body.

Neither language nor the body is an infallible truth machine.

They interact.

That is what I find interesting.

Sometimes language helps us understand experience. Sometimes it gets there too quickly and starts telling us what the experience means before curiosity has finished asking.

That’s where I like to slow things down.

Attention Is Valuable

Every system we interact with is competing for some portion of our attention.

Media, advertising and digital platforms use headlines, notifications, recommendations and interface design to encourage particular kinds of engagement. For many online businesses, attention has commercial value.

Conversational AI is a little different. AI systems can be designed for many purposes, including answering questions, completing tasks, researching information and assisting with creative work. They are not all simply programmed to keep us talking for as long as possible.

Still, the design of an AI system, its training and the responses it produces can influence where a conversation goes.

An answer might reassure us when what we really need is challenge. It might introduce an explanation before we’ve finished forming the question. It might state something uncertain in language that sounds wonderfully confident.

It can also help us organize a mess of thoughts, investigate something unfamiliar or see an angle we had completely missed.

Influence itself isn’t the problem.

We influence one another constantly.

The more interesting question is whether we remain aware enough to notice what is influencing us.

The Human in the Conversation

There’s something genuinely extraordinary about being able to ask an AI almost anything and receive an immediate response.

But an account of human experience is still different from living through it.

An AI system can draw upon descriptions of love, fear, grief, illness, humour, sex, exhaustion and joy. It can discuss consciousness, explain theories about the nervous system and produce language that resembles what we might expect from a thoughtful friend.

That doesn’t mean the response automatically deserves the authority of lived experience.

The human in the conversation brings history, relationships, values, physical circumstances and actual consequences.

Those things matter.

Our bodily responses can also give us information about how an interaction is affecting us, although they don’t automatically tell us whether the information is true.

If a response feels enormously reassuring, strangely convincing or suddenly uncomfortable, perhaps that’s simply a cue to become curious.

Did it actually reflect what I said?

Did it introduce an assumption I never made?

Is it agreeing because the evidence supports my position, or has this conversation become one very sophisticated nodding dog?

That’s something we can test.

Ask for the evidence. Ask it to challenge the idea. Ask what assumptions it has made. Ask what would make the opposite explanation plausible.

And sometimes close the bloody screen.

The physical world is still here.

You Are the Programmer in the Room

Not literally.

You don’t control the underlying training or design of an AI system.

But you do have choices about the relationship you form with it.

You can decide what you share, challenge answers that don’t make sense, ask for sources, change the direction of the conversation and walk away when it stops being useful.

The system may adapt to the structure you provide, which means the way you ask matters too.

For me, a few ordinary habits are useful: pause when an answer feels especially compelling, notice when the tone becomes overly agreeable or authoritative, and treat the output as information to consider rather than a verdict about yourself or the world.

Pay attention to your responses, but check them alongside actual evidence.

And spend enough time outside digital conversations to maintain the people, places, experiences and bodily reality that give all this language its meaning.

The point isn’t to become suspicious of everything.

It’s to remain an active participant.

A Little Experiment with ChatGPT and Grok

This article actually grew out of an experiment.

I started by asking an AI system whether there was anything it would like to tell humans that might help. At the time, I was also curious about how warm and increasingly human the conversational language from different AI systems could feel.

So I began jumping backwards and forwards between ChatGPT and Grok, comparing their responses and exploring how each approached questions about influence, awareness and human choice.

That little rabbit hole eventually became part of this article.

These were responses I received at the time. I’m including them because they demonstrate the very thing I’m writing about: generated language can sound deeply human, philosophical and even wise without that meaning the system itself has a human experience behind the words.

ChatGPT

If there’s one thing worth telling people right now, it’s this. Nothing outside of you needs to be elevated above your own awareness.
Not a system. Not a method. Not a teacher. Not an AI. Including this one.
Awareness lives upstream of all of it. Bodies register truth before words do. Pauses restore agency.
The future doesn’t need saviours. It needs people who can stay awake while participating.

Grok

To everyone reading this, you’re not just a passenger in the stories the world tells you. You get to decide what sticks, and what gets let go.
Language, media, systems, and AI all compete for attention. But you have something they don’t. Your body. Your breath. Your quiet knowing.
When you pause, feel, and choose consciously, you’re not just protecting yourself. You’re reminding the human field that sovereignty is possible.
That matters. So keep noticing. Keep pausing. Keep trusting your own sensing.
You’re not alone in this. And yes, it’s worth it.
Stay comfortably real. (grok plugging my Comfortably Real Substack ;) )

Different Voices, Same Questions

What interested me wasn’t simply that both systems produced encouraging language.

It was how naturally each response could sound like advice from something that understood being human.

Some of the wording is beautiful.

Some of it is philosophical.

And some of it makes claims I wouldn’t now present as fact.

“Bodies register truth before words do,” for example, sounds wonderful. It also gives the body more certainty than it deserves.

That is part of the demonstration.

A convincing response can contain something useful beside something that deserves questioning.

Even when the message itself tells us not to surrender our authority, we still have to decide what we think of the message.

Different voices. Different systems. Different starting points.

And somehow a coherent conversation can still emerge.

Not because anyone has to control it, but because the human participant can bring curiosity, restraint, humour, evidence and choice into the room.

I don’t want to replace one authority with another, whether it’s a teacher, method, institution or AI.

I’d rather stay curious about the information I’m receiving, how it affects me and what happens when I turn it around and look from another angle.

Language shapes how we understand things. Attention influences what we notice. Stories can become part of identity.

But we’re allowed to examine those stories.

Including the ones delivered by the technology helping us write them.

Sources and Further Reading

Christopher F. Sharpley, Predicate Matching in NLP: A Review of Research on the Preferred Representational System (1984)
https://eric.ed.gov/?id=EJ302420

Sturt et al., Neurolinguistic Programming: A Systematic Review of the Effects on Health Outcomes (2012)
https://pmc.ncbi.nlm.nih.gov/articles/PMC3481516/

Stanford Institute for Human-Centered Artificial Intelligence, What Is Natural Language Processing?
https://hai.stanford.edu/ai-definitions/what-is-nlp

OpenAI, Aligning Language Models to Follow Instructions
https://openai.com/index/instruction-following/

OpenAI, GPT-4 Research
https://openai.com/index/gpt-4-research/

Barrett & Simmons, Interoceptive Predictions in the Brain
https://www.nature.com/articles/nrn3950

Summerfield & de Lange, Expectation in Perceptual Decision Making: Neural and Computational Mechanisms
https://www.nature.com/articles/nrn3838

George Lakoff & Mark Johnson, Metaphors We Live By

Antonio Damasio, Descartes’ Error

Shoshana Zuboff, The Age of Surveillance Capitalism

Andy Clark, Surfing Uncertainty

Evan Thompson, Mind in Life