Is AI eroding slow mode? Push-button solutions versus thought partners, and approaches to challenging writing tasks

by Tom Johnson on Feb 9, 2025
categories: ai • writing

• Is AI eroding slow mode?
• Push-button solutions versus thought partners
• Experiment with using AI as a thought partner
• Approaching challenging documentation tasks
• Conclusion

Is AI eroding slow mode?

During the past few weeks, I’ve felt like my brain’s RPMs have been in the red zone. Granted, the constant stream of chaotic political news hasn’t helped—but regardless of current political events, I’m frequently checking the news, my email, and chat messages and operating in a mode that isn’t great. Reading long-form books has proven to be difficult. I run a book club at work focused on automotive and transportation books, and it took me two months to make it through a single book (granted, it was a 300-page historically dense book, but still). 

Last Friday, I started reevaluating my life and wondering about slow mode. I’d heard the term before but only had a vague idea of what it was, and I wondered if it could be an antidote to hyper-accelerated living. What would it take to operate in a way that’s more conducive to reading books, to adopt a pattern of thought that would allow me to participate in slower, more decelerated activities, like writing, that aren’t moving at chaotic speeds?

I have little doubt that, just like many other modern conveniences, AI tools contribute to fast mode. They foster short-form, consume-it-quick patterns of thought. When you use AI tools, especially to assist with writing, expecting quick outputs and responses becomes normal. For example, suppose you want to summarize a book. If you find the book’s PDF online, you can download it, upload it into an AI tool, and then ask it to read and summarize the book—all in minutes. Instead of drafting content from scratch, it’s easier to feed information into an AI tool and have it crank out an initial draft for you, getting you 80% of the way there in a matter of minutes.

In contrast, writing from scratch is a slow activity. (We can slow it down even further with long-form writing using a pen on paper, if desired.) Typing individual letters in each word slows down our thought processes and puts the brain in another mode. Typing out each word can seem painstakingly tedious compared to the rapid results with AI, where semi-intelligent words just appear magically on the page. It’s similar to the difference between microwaving a TV dinner and cooking it all from scratch. Who has an hour to spare when you can get a decent result in five minutes?

Is AI accelerating our brain’s CPU in a way that makes other activities feel impractically slow? Is it unbearable to sit down and spend an hour manually writing a blog post or manually chipping away at a long documentation project? It feels ridiculous even to say “an hour,” knowing that before AI, a blog post of 1,500 words or so would typically take me at least several hours, if not more.

Reading books is another activity that takes place in slow mode. It takes a special kind of patience to spend several minutes on a page, reading line after line, page after page, for upwards of an hour or more. Reading long-form books requires careful mental endurance—I’m starting to wonder if this patience for slow might be something I’ve lost. AI might have chipped away at this patience for reading and writing, along with all the other accelerated aspects of modern life.

I plan to investigate the slow movement a bit more. The slow movement started as a reaction against fast food and the ultra-processed outputs, the quick dinners on the go or in front of the TV instead of social meals with family and friends. But the slow movement has also spread to other aspects of life, beyond food. I’m not looking to become underproductive or sit on a couch staring at the wall, nor do I want to start cooking meals from scratch. But I do want to possess the ability to sit down and read a book for an hour without my mind jumping from task to task. I want to have the mental wherewithal to write a blog post for 2-3 hours without feeling that the task is impossible because it’s just taking too damn long.

On Friday night, I climbed into bed at around 7:30 pm and started reading for an hour until I was sleep-ready. Reading books, especially print books, has a soothing effect on my mind. When reading books, I slow down the pace at which my eyes move across the page. I no longer skim and scan like I do when reading online. I have to slow down, as if reading in slow motion intentionally. Then, when that pace becomes the new normal, I can absorb and enjoy the book. I’m not frenetically moving through large swaths of information, getting the gist of an article, and then moving on. That type of reading pattern makes it so I can barely read a paragraph in a book coherently.

I have no idea about the science, but it feels as though reading for extended periods of time generates different brain wavelengths. By the time I’m tired and ready for sleep, my brain is in a mode where it will sleep and stay asleep. 

I have another reason for investigating slow mode and reading, especially before sleep. For a couple of years, I’ve been waking up in the middle of the night (usually 3:30 am or 4:30 am) for inexplicable reasons, unable to return to sleep. My brain just wakes up thinking about this and that. My hope is that implementing the ritual of a lengthy reading sanctuary before bed will enable me to sleep peacefully through the night, without waking up randomly.

Expect more from me about slow mode. At this point in my journey, I’m just coming to grips with a few ideas:

• AI seems antithetical to slow mode.
• Writing and reading are slow activities.
• AI mindsets might be short-circuiting our patience for slow activities.

I’m also tinkering around with a theory about AI and creative writing. Some people claimed AI tools would amplify creative output, but I’ve been writing less on my blog since the advent of AI tools. I don’t know what to make of the decreased activity, but the fast mode versus slow mode might be a factor. Using AI tools to write more quickly at work, accelerating everything I do, perhaps persuaded me that writing manually on my blog is too tedious and time-consuming. The activity doesn’t fit into my accelerated pace.

But while AI tools do a decent job of creating technical documentation, they still fail at creative content (primarily for the reasons I explained in Seeing invisible details and avoiding predictable, conditioned thought). As a result, I don’t often use AI tools with blog posts except for proofreading. Overall, I’m producing less creative content on my blog because it feels so tedious to write, but I’m not turning to AI because I dislike its creative content output. Hence, I’m not writing many new posts at all. Have you noticed anything like this in your own creative endeavors? Is AI paradoxically stifling your creative output due to an accelerated mindset?

Push-button solutions versus thought partners

Let me address another thought I’ve had about AI tools. It seems that most AI writing tools today provide push-button commands, such as shortening content, creating a summary, converting items to a list, and so on.

While some of these commands might be useful, I prefer the free-form interface where I can ask the AI to perform my desired task. It’s much easier to type out these commands, such as “create a summary” of this content, or “convert this content to markdown,” etc., rather than looking for a specific button, clicking it, and not really knowing what prompt is being passed in the background.

If the result isn’t what I want (for example, if applying a style guide button changes too much of the text or reduces it unexpectedly in half), I can modify the prompt to say, “Only fix glaring grammatical errors, and don’t shorten the content.”

But my criticism goes deeper. Most conceptions about AI writing assistants focus on push-button commands to aid writing. Those push-button commands fall more into the simple writing tasks—the menial, mechanical tasks that don’t require much writing prowess or skill.

When I work on docs that do tap into my writing skills, the tasks are more like these: 

• Explain this extremely confusing concept
• Provide a sample workflow of how to use this API
• Describe what this complicated data means

There aren’t buttons for this. The workflow for using AI with these scenarios is entirely different from the push-button workflows. 

In The AI-Driven Leader: Harnessing AI to Make Faster, Smarter Decisions, Geoff Woods makes a point that resonates with me. Woods dismisses people who champion AI as a tool to quickly craft an email or perform other simplistic uses. That sort of task doesn’t move the needle. Using ChatGPT to draft a coherent email might save you 30 seconds here and there, but it’s not something that will help you develop a business strategy as a corporate leader.

Woods instead encourages readers to use AI as a “thought partner.” Use your AI thought partner to do tasks like these:

• Surface counterpoints to your ideas
• Challenge your assumptions
• Push your thinking in different directions
• Function as a sounding board
• Simulate potential scenarios
• Role play other points of view
• List arguments or considerations for decisions
• Analyze data and trends, and more

Overall, engage with AI as a thought partner to think more strategically. Don’t use AI for mechanical tasks alone.

When people consider prompt engineering techniques for docs, they too often gravitate toward push-button tasks—simplify this paragraph, format this content, or align it with our style guide. This short sells the power of AI.

It’s like having a legendary book editor, perhaps from a prominent New York publishing house, sitting next to you, and you ask them to perform trivial tasks on your content, such as shortening a paragraph here and there. No, you want this person to examine your book’s argument and analyze whether you’ve provided compelling logic. You want them to give insight into alternative approaches and how to incorporate storytelling elements. Don’t make AI just play the role of a copyeditor intern.

AI tools work pretty well as thought partners. Instead of using AI for push-button results, treat your AI like a critical thinker who can help you see through a problem. In the next section, I’ll provide an example.

Experiment with using AI as a thought partner

On a day when I was feeling a bit down, I listed 10 things that were bothering me. Then, one by one during the week, I tried to work through them using AI as a thought partner to see them in a different light, functioning partly as a noticing journal. A noticing journal prompts you to write down things you haven’t noticed before. Here’s my general prompt:

Prompt:

My list of negative items to work through included these:

• AI stock threatened by DeepSeek
• Tariffs risk economic chaos
• Feeling physically sick and congested

And about half a dozen more items!

Then, I started a conversation with AI about these topics. It didn’t take long before I began to see each of these negatives in a different light. For example:

• AI stock threatened by DeepSeek → If distillation makes the AI momentum fizzle, won’t that be great for writers anyway?
• Tariffs risk economic chaos → We’ll finally see if tariffs work, and if they don’t, political leaders will stop threatening everyone with them; they’ll have to political collaboration and compromise.
• Feeling physically sick and congested → This is an opportunity to see life from a different angle, to step back and experience the same events but another point of view, like taking off a mask.

I don’t need to get into the details. It’s something you can try. Think of something troubling you or something you’re depressed or upset about. Use it as a starting point for the dialogue. AI will transform into something much more than a copyediting intern for you.

Approaching challenging documentation tasks

Finally, in this collection of AI thoughts, I’d like to present a more concrete workflow for approaching more challenging documentation tasks with AI. As a technical writer, you know that there are often simple tasks (like updating a table with new properties, or indicating that a field is deprecated, etc.). Then, there are more challenging writing tasks that involve explaining complicated concepts. These are the tasks where you know you’ll have to roll up your sleeves and write a lot of content.

For example, engineers tell you that they’ve reworked how curvature works in the API and need you to explain all the new elements and how to use them, including their limitations and workflows.

For these more challenging tasks, I recommend this approach:

• Gather up the engineering diffs and reference documentation diffs available. Diffs are like magic for cutting through the difference between preliminary documents (such as engineering designers and one-pagers describing the product) versus what engineers actually coded and released. Use AI to generate readable summaries of these diffs.
• Gather other material related to the changes: engineering designs, one-pagers, partner bugs, and any other material that explains the features.
• Use AI to generate a first draft of the content, specifying what you need. For example, explain to the AI what your scenario is and what you need.

After you get that first draft, try layering on more information and detail.

This approach should get you about 75% of the way there. Of course, you’ll want to shape, edit, organize, and otherwise modify the content. But AI can help you get an early draft. You can take it from there yourself, or you can steer AI toward more edits.

When the content is 90% finalized, I recommend linking all the code elements to the regenerated reference documentation. This ensures that the new code elements actually exist. You might discover more questions during this phase, such as elements with highly similar names that could be confusing or inconsistencies in the implementation. Linking to elements will give your content a sense of authority and help users drill into the reference content for more details.

Conclusion

Granted, this last section seems to forget everthing I just argued about AI eroding slow mode. It’s like wait a minute, wasn’t Tom just arguing for doing things, such as writing and reading, the manual way as a strategy for recapturing slowness into his life? Well, yes, but I’m not ready to embrace slow mode with documentation. I’m not entirely sure why. It could be that I care very little about the ACME method—the faster I get done with writing about the ACME method, the more quickly I can return to my decelerated activities.