When I sat down to write this post, I wrote :

Write an introductory paragraph telling a story about when I was in grade school, my 5th grade teacher, Mrs. L, insisted on the class outlining their paper.

Write a paragraph about how we all wish we would outline, & there are frameworks like Situation, Complication, Question, & Answer that achieve this goal. But it’s been easier to simply dive in as much for grade school as in corporate life.

Write a paragraph that today, with AI, writing has become easier because the outline is (i) essential to prompting AI to steer the narrative ; (ii) it’s easier to edit than it is to write ; (iii) incomplete thoughts can populate the outline & AI will interpolate to fit them in.

Those are AI prompts. Within 45 seconds after drafting that outline, AI could have written a full post on the topic. In fact, it did & it’s here, a bit flowery & verbose, but on point.

It’s no different in software. When I wrote analysis code two years ago, I typed every line.

Today, my analysis code starts like this:

# read the CSV file financings.csv into a dataframe
# convert the Date column to date format
# summarize the data by summing the financing by date
# plot the data on a line chart with the line thickness 2 in orange

The computer fills in the code afterwards. Most of the time I execute the R script without changing code.

In computer science, this is called metaprogramming.

I suspect we’ll call the first section of the blog post metawriting.

This type of meta-activity is the reason AI will massively increase our productivity. AI frees us to focus on the higher-order goal while the computer sweats the details.

What hasn’t changed about writing since the dawn of AI?

A good outline remains essential. Grammar teachers rejoice!