Submitting AI-Generated Outputs as They Are

Summary

• The "paradigms of format" for materials are as follows:
  • 1: Formal. For example, formal PowerPoint presentations.
  • 2: Rough. For instance, notes written in note-taking apps, wikis, or Markdown.
  • 3: Mechanical. Such as reports or conversation logs generated by AI.
• Depending on the context, you can elevate the paradigm if it is an internal submission.
  • Particularly, it's beneficial to adapt to the third paradigm, Mechanical.
• Shifting to the Mechanical paradigm can speed up the creation and submission of materials, thereby accelerating decision-making processes as well.

Background

Content Over Format

We often focus on the format of the materials for no apparent reason. We tend to embellish what could be just five bullet points of plain text. This often results in the voluminous and noisy documents or slides you may be familiar with.

Recently, there has been a pattern of completing reading and writing solely through a browser—submitting rough notes via pages like notes, wikis, or tickets is becoming more common, but there are still many places where this doesn't hold. For instance, when presenting to managers or clients, a formal format is often required. Failing to produce a formal format can sometimes lead to being deemed incompetent.

Depending on the context, unless it's a formal submission or presentation, the format is often excessive. The key point is the content; as long as the core message is conveyed, that's all that matters.

Paradigms of Format

Formats of materials also have paradigms.

The first paradigm is the Formal Style, which includes formal PowerPoints or documents.

The second paradigm is the Rough Style, such as notes written in apps, wikis, or in Markdown. For convenience, it's best if those can be read as HTML from a browser and each page has its own URL. Among familiar colleagues like engineers, even Markdown files can work.

But paradigms don't stop here.

Mechanical Style

Mechanical Style is the third paradigm of material format and includes outputs from generative AI or conversation logs.

Creating materials in the Mechanical Style essentially means using AI-generated outputs as they are. You submit the output directly to your team, superiors, or clients.

Think of it more as submitting the conversation log with the AI. Not just the final output but also the instructions that led to its creation are part of the material.

Benefits

By adopting the Mechanical Style, you can accelerate decision-making processes using the materials.

Q&A

Here are answers to frequently asked questions.

Q: Aren't AI-generated results unreliable?

Ans: Human results can be unreliable too.

In general, there are few scenarios where you must absolutely trust the results, whether they're from humans or AI. If the context does require such trust, just incorporate human verification or refinement processes.

We are engineers and desk workers. Most of our work involves wisely using "hints in the form of information." Information is just a hint; its accuracy or quality isn't always crucial. What's more important is the quantity and diversity (variety and richness) of hints and most importantly, us making concrete decisions based on them.

You can quickly produce materials of adequate quality and submit them. The readers, too, should quickly read and use the materials with the notion that they are mere hints. Except for its lack of precision, AI has surpassed humans, and in terms of creating hints, humans alone can't compete. It's far quicker and higher quality to collaborate with AI than it is to produce something on your own.

Q: What is needed to read Mechanical Style materials?

Ans: Decision-making and adaptation.

Firstly, you need decision-making. It's premised on the choice to use "hints in the form of information" to make your own decisions and take action instead of being an evasive person who only does what they are told. Those unable to make decisions won't be able to read Mechanical Style materials, and even if they do, they'll only be confused, wondering, "So, what should I do?"

Such individuals (I call them Reactive persons) are convenient for doing simple tasks as ordered, but they're otherwise ineffective and likely won't move past the first paradigm. It's a waste to deal with them; either follow the first paradigm quietly or leave if you truly dislike it.

After that, it's about getting used to it. As a knowledge architect, I can almost assert that we are more bound to format than we think. Even if something can be expressed in five bullet points, we tend to not accept it and prefer to expand it into rich content many times over. This applies even to rational executives.

That's precisely why we need to consciously engage with materials in the Mechanical Style and get used to them. It's a simple matter of increasing your experience using generative AI. Tools like ChatGPT's chat interface are perfectly fine. One benchmark could be whether you query the AI more often than a search engine. If your mental model has evolved to that point, then you are likely accustomed to the Mechanical Style.

Q: Documents created in Mechanical Style are lengthy and difficult to read.

Ans: Improve your prompts.

The issue is not with the Mechanical Style itself but with the producer (the prompt engineering). It's merely a problem with how the material is created, so it's a call to change the way you're doing it, learn, and enhance your skills.