Garden variety
Every spring, I plant tomatoes and whatever else looks good at the plant nursery. I do not have a green thumb, so I am inordinately proud when I harvest food that my family is able to eat.
We also have a farm box delivered to our house weekly all summer. It is highly convenient, and the produce is better than anything I grow myself. The other day my daughter asked me why I plant my own garden when the veggie box is delicious and comes every week. It's a good question.
Why do I keep on planting tomatoes?
How does your garden grow?
I may grow my own tomatoes, but there are plenty of places where I take the easy road when it's available. I buy clothes rather than sewing them myself. I use computers to write practice math problems for my daughter rather than writing new questions by hand. I have my other daughter's volleyball shoes delivered rather than driving to the faraway specialty store to pick them up.
If you're like me, there are actually very few places where you choose to do something manually when a fast, easy, and reasonably affordable alternative is available. Which makes the exceptions extra interesting.
Last year, I wrote about the places I use AI in the making of this newsletter. Finding patterns in my engagement data and offering insights about my audience demographics? Yes, absolutely. Writing newsletter copy, building data visualizations, and drawing comics? No way.
I have the skills to analyze my engagement data by hand, but I choose to use AI instead. My personal effort on this task would be neither valuable nor enjoyable enough to justify the time spent. On the other hand, I choose to build my newsletter content the old-fashioned way because doing it myself is how I form my POV about topics that are important to me.
When your so-so effort is worth it
In a world where AI can do a so-so job with most knowledge work, it's fascinating that any of us hang on to doing any knowledge work manually. Part of this is our natural aversion to change; by default, we like to do things the way we've always done them, because we know that it works. But I don't think this is the whole story.
A few weeks ago, I also shared data showing that people are less likely to use AI in areas where they are experts. While this is an important consideration, it doesn't explain all the patterns. I draw my own comics not because I'm an excellent visual artist (I am not), but for the same reason I grow my own tomatoes: because it is satisfying and fun.
If you reflect on the places where you could choose to use AI but choose not to, I suspect they fall into four categories:
- Your work is better than what AI would produce, and the stakes are important enough for this to matter
- The task is educational or fun for you; even if AI could do it just as well or better, you do it yourself for the sake of learning or personal fulfillment
- You see no reason to change; things seem to be working fine as they are
- You object to the use of AI, either in the particular situation or more broadly
Model maximalists would argue that (1) is a temporary state. Eventually, they believe, models will become powerful enough that they can do nearly every knowledge task as well or better than people can. Model maximalists would also claim that (3) and (4) are short-term considerations in the face of increasingly powerful models. The argument is that someone will eventually use AI for those scenarios and demonstrate what is possible. I mostly believe these arguments.
But even the most die-hard model apologist can't argue with (2). Because for the person choosing the harder path, the decision is not strictly rational. It is not about efficiency or quality as the top goal. It's about learning, fun, and personal fulfillment.
It might not make sense to grow my own tomatoes. But I'm still doing it anyway.
Kieran
Want to get better at data storytelling? I've got you covered.
If you liked this story, why not subscribe to nerd processor and get the back issues?
My latest data stories | Build like a founder | nerdprocessor.com