My Year in Poetry

In 2024, I had a near-daily habit of writing morning pages. I wrote a total of 212 entries totaling to 98K words. For that amount of content, I was sure I could find some interesting themes in there.

I wanted to abstract the entries into a higher-dimensionsal space so they could easily be clustered. But, what should they be grouped by? Mood, emotion, the topic being discussed?

I wanted something broader to capture where my head was steeped, and landed on what are the questions that I'm trying to answer? After all, life can be seen as a series of questions that we're trying to answer, for example:

• How can I make this relationship better?
• How can I show up for my friends in a way that feels authentic?
• Should I go on vacation or focus on the product launch?
• Why am I feeling down?

For each journal entry I asked the AI: what question is the author implicitly trying to answer? To abstract even further, I asked it to broaden the question (remove any specifics), then go even further by providing a larger philosophical question being answered.

From there, I created embeddings for each of the broader questions and reduced them to a 3-dimensional space and plotted them. You could start to see organic clusters form based on the topics and nature of questions I was answering. Entire clusters were based on my relationship at the time, the company I was starting, working with my family, or thinking about community and loneliness after moving to a new city.

Each of those clusters were given a distinct color below:

For each cluster, I took all the questions & implied answers and sent them back to the AI. This time I asked what is the broad piercing question that ties all these questions together?

What I got back was a set of ~10 or so questions that I kept circling around through the year. A particularly resonant one was:

How do I honor both my own needs and my responsibilities to others without losing myself?

I was a shook by how specific and relevant the questions were. 2024 was a particularly tough year, and this distilled it all down. It was both relieving and painful to read - on one hand I felt seen and on the other I was indignant because the whole story wasn't captured in these journals. I wanted it to know more.

I then asked it to give me a summary of my answers to the questions, and a set of recommendations from a therapist/coach for each big question I'm pondering. From there, outline what a day of my life looks like today VS after I take the recommendations.

After doing this for all 10 questions, I handed the full analyses to Claude again and asked it to create a poem in the style of Khalil Gibran. A poem that I could read every morning. It responded with this beauty -

Morning Invocation

And so, as the sun greets another day, I speak these truths to my awakening spirit:

I am not broken seeking wholeness, but whole seeking expression. I am not lost seeking direction, but exploring seeking wisdom. I am not empty seeking fulfillment, but full seeking flow. I am not alone seeking connection, but connected seeking depth.

Let me move through this day as water moves through the river's course - flowing with purpose, yielding without losing direction, powerful in gentleness, clear in nature, and ever true to my own depths.

For in this journey of becoming, I am already complete. In this dance of growth, I am already whole. In this quest for connection, I am already home. In this search for meaning, I am already found.

And with each breath, I remember: I am the author of my story, The keeper of my truth, The holder of my power, And the vessel of possibility.

So let it be.

Giving AI a Soul

LLMs are just predictive algorithms. They use your prompt to launch into a probability space with the average of everything that has ever been said. If you give it average input, then it will provide average output.

However, if you give it deeply human inputs, then it'll narrow it's probability space and provide you with deeply human outputs. Perhaps similarly to how humans work - if you open up to another person, then they're more likely to open up to you.

Maybe that's how we can relate with AI as a whole. Instead of relying on it to become superhuman, we can rely on it to reflect back our own humanity. This poem is a reflection of my own thoughts, except amplified with the wisdom of others. Maybe that's what it means to be superhuman.