How much does AI sound like me?

I try to use AI sparingly when writing. One thing I avoid is using AI to write for me. We all know how AI slop sounds. Or do we?

With Claude AI, and I'm sure others, you can feed it pieces of your writing to have it write in your voice. I was curious so I gave Claude my last 6 newsletters and had it write a few paragraphs. What I want to find out is how close AI can get to my voice.

Below are two snippets about shifting from external to internal validation - something I've been thinking about a lot recently.

One of these snippets is written by me and the other by AI. Can you guess which one?

Snippet One

What if you couldn't tell anyone about your achievements? Would you still pursue them with the same passion? That's a powerful question to consider when examining our relationship with validation.

The shift from external to internal validation begins when we start measuring our work against our own standards rather than anticipating others' reactions. It involves creating personal benchmarks that reflect what truly matters to us—not what might generate likes, praise, or recognition.

When we design our projects, write our code, or craft our words primarily to satisfy our own curiosity and values, we cultivate a resilient sense of accomplishment that doesn't evaporate when external feedback is mixed or absent.

This transition isn't about rejecting feedback entirely—it's about ensuring your motivation comes from within first, allowing external validation to serve as a pleasant bonus rather than the driving force behind your efforts. The most fulfilling work often happens in those quiet moments when you're completely absorbed in the process, regardless of who might eventually see the results.

Snippet Two

Would you still do what you do if nobody sees it? A question around external validation that I would hope to answer yes. But the allure of external validation is powerful. And social media has made it all the more enticing and destructive.

Something I've observed in myself is the inability to separate metrics from my own self worth. For example, if I publish a newsletter and I see a few people unsubscribe. It tells me that my last newsletter was unworthy. That I've failed.

External validation causes us to make opinionated choices for the wrong reasons. Either we change to get more likes. Or we become discouraged and give up. What we need to develop instead is internal validation.

We can do that by fulfilling three needs — autonomy, competence, and relatedness.

These three needs encompass self-determination theory which posits that people are more motivated when they feel they have ownership over their lives, mastery of skills, and deep relationships.

Conclusion

If you thought the first snippet was written by AI, you're right!

It's unsettling how close it's getting. Especially, how both snippets started with a similar question. But the AI snippet has a very uniform writing rhythm - long sentences about the same length.

Maybe if I gave it more examples it would get closer and closer to my voice. I could also see using AI to generate text which you would edit to fit your voice. It would make the writing process faster.

But I don't know if we want that. I certainly don't.

I don't want AI to pigeonhole my writing style. Maybe. I. Want. To. Have. One. Word. Sentences. Everywhere.

I also want to struggle with writing. It ties into the thought on validation. Building competency is how we find internal validation. I don't want AI to take that journey away from me.

A bit of a longer newsletter today, but an experiment I was curious to try. Onto the links!

Cool stuff around the web

Kent Beck writes about Augmenting Coding & Design - Development feels like breathing but AI is a genie that just inhales. It continually adds features increasing complexity. How can we use AI to breathe - not only inahale?

Sink is a self hosted link shortener with analytics - I'm taking notes on what needs to go into a self hosted open source project.

Sam Rose wrote about Reservoir Sampling - I wish all algorithm explanations were as interactive as this post.

Thanks for reading this week's newsletter. If you have any thoughts or questions, feel free to reach out by replying to this email or send me a DM on Bluesky.

Catch you next week!

Jono