Learn AI With Me: This Is Programming

Day 12, in which I consider what it means to write a prompt.

Yesterday, I shared the frustration I experienced when I was trying to publish a set of links about AI, using a variety of AI tools to help with that process. In the end, I achieved the bare minimum. But the process also confirmed for me something about AI.

Using AI well, at least at this moment, resembles computer programming. In fact, the more you ask AI to do, the more like computer programming this process becomes.

Consider this prompt

Yesterday, I wrote this prompt for Claude AI:

I have attached a text file with a list of links. I would like you to format this into a table with the following columns: Column 1: Title
Column 2: URL
Column 3: Description
Column 4: Category

Please parse the text file so that each row of the resulting table contains exactly one URL in column 2.

Here are the definitions of each column:
Column 1: Title. If a title is available, please put it in this column. If it's not available, leave the column blank for this row.
Column 2: The URL
Column 3: If a description is available in the text file, insert it here. If not, follow the URL and extract a short (250 character) description from what you find.
Column 4: Category, if a description is available, read the category, then assign one of the following categories: [Getting Started], [Experience Report], [Ethics], [Economics], [Environment], [Resources], [Smart People To Follow], [Tools]

Please return the results with both a partial screen preview as well as a .CSV file that contains the complete results.

This is clearly a computer program, albeit a simple one, written in English.

What Is Programming?

According to Wikipedia:

Computer programming or coding is the composition of sequences of instructions, called programs, that computers can follow to perform tasks.

And what about Google?

Computer programming is the process of writing step-by-step instructions, or code, in a specific programming language that tells a computer how to perform a task or solve a problem.

What about Programming Languages, or “code” as we love to say these days? Well, Wikipedia says:

Programmers typically use high-level programming languages that are more easily intelligible to humans than machine code, which is directly executed by the central processing unit.

I think you’d have to adopt an extremely narrow point-of-view to consider my prompt anything other than “programming.” Clearly, a prompt is, or can be, a program.

This is not a unique insight. In fact, people operating on this idea often use the phrase “prompt engineering” to describe what they’re doing. But do we really need another term here?

Who Cares If It’s Programming?

So much of the conversation about AI has been about economic disruption. Will it put developers out of work? Will it put designers out of work. What about the rest of us?

Clearly no. Whether we’re writing prompts and submitting them to LLMs or writing in structured, special-purpose programming languages, AI isn’t going to remove the need for software programmers. Sure, it’ll change their workflow, and perhaps extend programming capabilities to non-specialists—like me. But the work—the thinking about structure, process, flow, architecture—this work remains.

And what about designers?
Clearly, my experience with Notion yesterday demonstrates the technology industry’s urgent need to make better use of design. Notion is a tool that I was just begging to use, and they were just begging to take my money. But the utterly baffling user interface that I encountered was a barrier that prevented both of us from getting what we want.

AI-enabled tools are not going to win without design.