I fed my entire life story to 4 different AIs to see if they’d tell me I was full of shit

I’ve been reading Tony Robbins’ Awake The Giant Within.

Yeah I know. Self-help book…
Well, actually one of the best book I’ve read for years so you should give it a try before judging ;)

Two ideas in there actually stopped me mid-page :
Patterns of behavior and identity modeling.

The argument is brutally simple :
You are the sum of your behavior. (Alex Hormozi has a really similar mindset)
Not what you think you are. Not what you tell people at parties. What you actually do, repeated over time.

And if your past behavior doesn’t point in the direction of the identity you think you have ? You’ve got a problem.
Or worse - sometimes your patterns are exactly telling you who you are.
And that person might just be a big pile of shit.

That’s life, we all start somewhere.

So reading this, I tought : “there’s something to test here with AI”.

So I ran a weird experiment on myself.

The setup

First, I wrote my own past, as detailed as I could.

Couple of pages with no interpretation.
Just raw memories, juxtaposed with real dates : video releases, major life events nd whatever I could anchor in time to keep it honest. (because first thing I saw is that my memories were completely out of sync with the real timed-events)

Then I analyzed myself from that. (Alone with no AI)
What are my behavior patterns. Which ones are constructive or actively destructive.

And then - here’s the fun part - I sent that exact same detailed past to 4 different top-tier AI models (Claude, GPT, Gemini, Grok) and asked them to do the exact same analysis I just did.

What I expected

Two possible outcomes going in :

Either the 4 models spit out wildly different analysis, which would basically prove they’re just fancy noise generators with no actual signal.

Or they’d converge on something, but bring angles I completely missed because they don’t have 30 years of my dumb ego protecting me from certain conclusions.

I was actually betting on the first one haha. (The ego is a bitch)

What actually happened

First thing that hit me : doing my own analysis before touching any AI helped tremendously.

Because I have context in my head that’s basically impossible to transcribe into text. Some memories carry weight that no amount of detailed writing can fully convey to something that’s never lived a single day.
My own insights ended up more specific and more true (to me) than anything the models produced.

That said, the AI output did surface a few things I’d missed. Genuinely useful angles.
Roughly 60% of what came back matched what I’d already found on my own, 30% was just misinterpretations and 10% were actual angles I’ve completely missed.

And here’s the part that actually surprised me :

There was basically no difference in content between the 4 models.

Claude, GPT, Gemini, Grok.
Same core patterns identified. Same structure. Same conclusions, more or less.

The real differences were in tone and delivery, not substance.
Grok, for the record, was by far the most painful to read because of its lack of structure in its writing.

GPT and Claude had maybe 2-3 more diverse insights that Gemini and Grok didn’t touch. But that’s it. Not a chasm. A small gap.

(good to know when you know the difference in price/token haha)

The reassuring part

I learned less than I expected from the AI side of this. Which, weirdly, reassured the hell out of me.

Going in, I thought I’d see a massive gap between my own interpretation and the AI’s - because of my biases, my blind spots, the stuff I unconsciously protect myself from seeing.

Instead ? Pretty damn close. (And I often ended up judging myself more harshly than the AIs - but this could come from the tone I use to write my past so this should be taken carefully)

Which tells me something a lot of people don’t want to hear right now :
AI is not that smart.

Not in the “understanding a human life and drawing real insight from lived experience” sense anyway.
It’s good. It’s useful. It caught things. But it’s not some oracle seeing through your soul better than you can see through it yourself.
AND, its usefulness was 10-20% not 90% of the actual work.

And honestly ? That gives me a bit of hope for humanity’s usefulness going forward.
We’re not as replaceable as the hype wants us to believe.

The actual point of all this

Here’s the part that matters more than the AI comparison itself.

From my own analysis, combined with what the models added, I had something I didn’t have before :
A clear, evidenced list of which patterns in my life are constructive and which ones are actively working against the person I want to become.

And from that clarity, I could actually decide to break some of the destructive ones.
Not vaguely “be better”. Specifically : this pattern, repeated on these dates, led to this outcome, stop doing this exact thing.

We’ll see how it goes !
You can ask me in a year to see if I bullshited myself lol.

What’s next

I’m seriously considering building a small piece of software around this exact process.
Something that lets anyone do this full life balance sheet: write your past, analyze your patterns, cross-check with AI, and walk out the other side with an actual decision on who you want to become next.

If you want the details on the exact process I used, or you’re curious about the prompts I sent to each model, shoot me an email (romain.gilliot@motion-live.com).

And if this software actually happens, you’ll hear about it here first.

Stay tuned.