I Stopped Using AI To Code For 30 Days

What I Learned in 30 Days of Coding Without AI

Introduction to the Challenge

• The speaker shares their experience over the past 30 days, focusing on building apps, writing scripts, debugging, and solving technical issues without using AI.

• They reference a previous video titled "AI coding sucks," where they expressed their frustrations with AI in coding and announced a 30-day break from it.

Reflection on AI Usage

• The speaker invites viewers to share their experiences regarding changes in workflows or breaks from AI after watching the previous video.

• They introduce themselves as CJ and set the stage for discussing their coding journey without AI.

First Day of Coding Without AI

• On the first day, CJ attempts to reverse engineer how to order pizza through a specific website as part of a challenge with Scott Wes.

• They describe feeling urges to use AI while coding, likening it to an addiction when trying to quit cold turkey.

Challenges Faced During Coding

• The task involved replicating requests made by the website's GraphQL API, which became repetitive and tedious.

• CJ reflects that had they completed several functions first, they could have provided examples for AI to learn from and improve its output consistency.

Key Takeaways from Coding Experience

• Writing core logic before involving AI can enhance its performance; providing structured examples helps achieve better results.

• Each function followed a similar process: determining JSON payload requirements, crafting GraphQL queries, and ensuring type safety with TypeScript types.

Embracing Creativity Beyond Coding

• In addition to coding, CJ explored photography during this period. This creative outlet provided enjoyment outside of programming.

• They emphasize the importance of engaging in diverse creative activities rather than solely relying on coding for fulfillment.

Observations on Content Creation Trends

• CJ expresses concern about the increasing prevalence of AI-generated content in social media and aims to create authentic human-generated content instead.

Presentation Preparation for Honom Conference

Image Downloading Script and Documentation Practices

Creating a Node.js Script for Image Downloads

• The speaker discusses the need to download royalty-free images from Pixels while ensuring proper credit to authors. They emphasize the importance of offline accessibility for presentations.

• A simple Node.js script named "get background" is introduced, which reads arguments using process.argv, highlighting the necessity of understanding this aspect as a Node developer.

• The speaker reflects on their reliance on Stack Overflow for coding solutions but expresses a shift towards consulting official documentation instead, influenced by AI-generated code similarities.

• Devdocs is mentioned as a valuable resource that provides offline access to various documentation, allowing developers to search effectively without internet connectivity.

• The script's functionality includes downloading an image from a URL and saving it locally, with metadata management for easy reference in slides.

Emphasizing Documentation Over Community Resources

• The speaker notes that AI's increasing use may lead to fewer questions being asked on platforms like Stack Overflow, making documentation skills more critical for developers.

• They stress the importance of practicing how to read and search through documentation effectively, especially when integrating AI into workflows.

Game Development for GitHub Universe Booth

Building "Spot the Syntax Error" Game

• The next project involves creating an interactive game called "Spot the Syntax Error" for Sentry and Syntax at GitHub Universe, designed to engage attendees with fun challenges.

• Players scan their badges to start playing; they must identify syntax errors within JavaScript code snippets presented during timed rounds.

• Real-time leaderboard functionality is implemented so players can see their rankings immediately after answering questions correctly.

Technical Considerations in Game Development

• Data storage decisions are crucial; user responses are tracked in a database while preventing unauthorized access to answers through backend validation mechanisms.

Building an Application: Choosing Technology

The Process of Selecting Technology

• The speaker discusses the manual process of selecting technology for building an application, which took about one to two hours. They ultimately chose Firebase Realtime over Firestore due to specific reasons.

• Emphasizes that while AI can assist in tech selection, it should not replace human decision-making. AI could perform tech spikes and provide analysis but shouldn't take full control.

• Reflects on their experience as a consultant where they assessed app ideas for technical feasibility, technology stack, timelines, and unknown factors—skills that contribute to their senior developer status.

Transitioning to Writing Projects

• The speaker shifts focus from coding to installing and configuring a Linux system for writing purposes during November, known as National Novel Writing Month (NaNoWriMo).

Creating a Distraction-Free Writing Environment

Concept of Writer Decks

• Introduces the concept of "writer decks," designed for distraction-free writing by minimizing interruptions from multi-purpose devices like laptops and phones.

• Highlights the appeal of building these devices using old hardware such as laptops or Raspberry Pis, creating a unique environment conducive to focused writing.

Setting Up the Device

• Describes repurposing an old netbook from 2008 into a distraction-free drafting device by booting directly into terminal mode without a graphical user interface.

• Utilizes Micro, a simplified text editor similar to Nano, ensuring no distractions like video streaming are possible on this setup.

Challenges Faced During Setup

Technical Hurdles

• Discusses challenges encountered due to the netbook's 32-bit processor while most Linux distributions have transitioned to 64-bit systems; found Debian Bookworm version 12 supports 32-bit.

Features Implemented

• Wanted features included automatic login upon startup and sleep mode when closing the lid. Additionally set up auto-saving every minute with Git integration for version control.

Limitations of AI Assistance

Research Challenges

• Shares experiences searching online for solutions regarding Linux commands; highlights instances where AI tools provided incorrect information despite appearing credible.

Conclusion on AI Dependency

The Impact of AI on Information Quality and Learning

Concerns About AI-Generated Content

• The speaker expresses concern over the prevalence of AI-generated content, noting that many articles published before 2021 contain valuable information compared to newer AI-sourced material.

• A specific example is given regarding Linux.com, which allegedly uses AI to generate articles by scraping top search results, leading to potentially misleading or inaccurate information.

• The speaker highlights a cycle where AI-generated answers reference other AI-generated content, resulting in a compounding issue of misinformation online.

Reflections on Learning and Expertise

• The speaker reflects on the shift from pre-AI internet resources being rich in quality information to a landscape dominated by less reliable, AI-produced content.

• There is a suggestion that future searches may need to exclude results from after 2021 or 2022 due to the overwhelming influence of self-referential AI outputs.

Recommendations for Future Engagement with AI

• Moving forward, the speaker advocates for individuals to conduct their own research and establish foundational knowledge before integrating AI tools into their workflow.