DEV Community

Alijah Konikowski
Alijah Konikowski

Posted on

From My Experience: Does AI Make Me Work Less, More, or About the Same?

#from #your #work #less

From My Experience: Does AI Make Me Work Less, More, or About the Same?

Introduction

As a technical writer, I’ve been experimenting with AI tools – primarily large language models (LLMs) like GPT-4 – for several months now. Initially, I approached it with a healthy dose of skepticism. Could a machine really help me write more efficiently? The short answer is: it’s complicated. It hasn’t dramatically reduced my workload, but it’s fundamentally shifted how I work, creating a noticeable increase in output and a slightly different kind of cognitive load. It’s not less work, but it’s demonstrably different work. This article will detail my experience, breaking down how AI has impacted my daily tasks and offering a realistic perspective for developers and other knowledge workers.

Core Concepts

The key isn't that AI is a replacement for human creativity or critical thinking. It's a powerful augmentation tool. Here's a breakdown of the ways AI has affected my workflow:

  • Reduced Initial Draft Time: Previously, starting a technical article required significant research, outlining, and generating a rough first draft. AI excels at generating these initial drafts based on a prompt. I’ve found that providing a concise prompt outlining the topic, target audience, and desired tone can produce a usable draft in minutes, compared to potentially hours.
  • Enhanced Research: AI can quickly synthesize information from multiple sources. Instead of manually searching through documentation, blogs, and articles, I can ask it to summarize key concepts or identify relevant examples.
  • Improved Clarity and Conciseness: AI can help refine writing, identifying areas where phrasing is clunky or overly verbose. I use it to rephrase sentences and paragraphs, ensuring clarity and readability.
  • Shifted Focus – Editing & Validation: The biggest change is that the bulk of the production time is now dedicated to editing, fact-checking, and validating the AI-generated content. The AI provides a starting point; my role is to polish and ensure accuracy.
  • Cognitive Overhead Adjustment: Initially, there was a significant cognitive load associated with prompt engineering – figuring out how to phrase requests to get the desired output. I’ve become better at crafting effective prompts over time, reducing this initial overhead.

Practical Example

Let’s consider a recent article I needed to write on "Implementing OAuth 2.0 in a Node.js Application." Without AI, this would have involved:

  1. Research: Spending 2-3 hours researching OAuth 2.0 concepts, libraries, and best practices.
  2. Outline: Creating a detailed outline with sections on registration, authorization, token exchange, and refresh tokens.
  3. Drafting: Writing approximately 3-4 hours of content, including explanations, code snippets, and example scenarios.
  4. Review & Editing: 1-2 hours refining the content for clarity, consistency, and technical accuracy.

Using an LLM, this process changed to something like this:

  1. Prompt: “Write a technical article explaining how to implement OAuth 2.0 using Passport.js in a Node.js application. Target audience: JavaScript developers with intermediate experience. Include sections on registering an application, handling authorization codes, exchanging authorization codes for access tokens, and refreshing access tokens. Include a code snippet demonstrating token exchange.”
  2. AI Generation: The AI produced a roughly 700-word draft in 15-20 minutes.
  3. Editing & Validation: I spent approximately 1.5 - 2 hours reviewing the draft, adding clarifying comments, correcting minor errors, ensuring the code snippet was accurate and up-to-date (using npm outdated), and adding a brief conclusion.

The total time investment was reduced from roughly 6-8 hours to approximately 3.5 hours. Crucially, the quality of the initial draft was high enough that it served as an excellent foundation for my edits.

// Example Token Exchange (Simplified - Requires actual implementation)
// This is just to show the conceptual flow.
const accessToken = await exchangeCode(authorizationCode, clientSecret);
Enter fullscreen mode Exit fullscreen mode

Conclusion

AI isn’t a silver bullet that eliminates work. It’s a tool that fundamentally alters the nature of the work. I'm not working less overall, but I’m working differently. My time is now spent more strategically – on tasks requiring higher-level cognitive skills like critical thinking, validation, and ensuring accuracy – rather than on repetitive, lower-value activities like generating initial drafts.

The initial learning curve associated with prompt engineering is real, and the responsibility for ensuring the AI-generated content is accurate and reliable rests squarely on the human user. However, for knowledge workers like technical writers, the potential benefits in terms of increased efficiency and output are substantial. As AI models continue to evolve, I anticipate this shift in workload will only become more pronounced, requiring a proactive approach to learning how to effectively leverage these tools and adapt our workflows accordingly. The future of writing, and likely many other professions, isn't about being replaced by AI, but about learning to collaborate with it effectively.

Top comments (0)