Blog Post by Yari
October 18th, 2025
Today we’ll explore the current controversy surrounding the use of vibe coding and Large Language Models (LLMs) like ChatGPT. We’ll acknowledge the pros and cons of these tools and emphasize the importance of critical thinking and due diligence when using them.
But we are professionals, and pros always aim to effectively leverage while mitigating any risk.
Girl code translation: If you learn to use something, you lower your risk of fallback. It’s like learning that Louboutin tends to run a half size smaller, so you size up next time. By doing that, you have just lowered the risk of pain and mitigated risk.
The use of LLMs and vibe coding techniques is a hot topic.
Some people love it, while others are skeptical, arguing that it diminishes critical thinking and make us both lazy and boring. This author (Bridgerton reference) can see both sides of the argument, but if you must do it responsibly, you don’t want to be that classmate who uses ChatGPT and then denies it, even though you both know the truth about why there’s 26 em dashes in that document.
But alas, the reality is that the world is changing, and many companies are already adopting LLMs. The AI bubble is something we’ll talk about before we move on to complex modeling. So, it is crucial to familiarize yourself with these technologies because if you don’t, you’ll be left behind.
Meet Your New Intern: The Newest Pop Star Lineup, LLM ChatGPT
If you ever have to use an LLM, and you most likely will, it’s best to treat it like an intern. You want to get to know them, familiarize yourself with their capabilities, but you should always double check their work. LLMs like ChatGPT are prone to hallucinations. This doesn’t mean they’re having a DSM level mental breakdown or entering psychosis. It simply means that the output you receive might be factually incorrect.
It’s like an intern making a mistake that could reflect poorly on the company, as was the case with a certain firm and the Australian government this month. So, you, as the professional, need to catch these errors before they lead to embarrassment and mistrust. You never want a real life 404 error, or a technical error.
Here are some rules to abide by:
Rule #1: Double Check, Then Double Check Your Double Check
In corporate, we call this due diligence.
The first and most important rule is to always double check the work, whether it’s from an LLM or from any code that generates an output, in coding your output is your code return. Always double check your LLM intern. Make it a habit to verify the information you are getting from any source. Lies and misinformation are everywhere, and lies are like rumors, they are rarely rooted in truth. Once they seep in, suddenly your group isn’t as friendly. You’ll always want to avoid this real life 404 error.
LLMs struggle with interpreting ambiguity at times. We’ll go into more detail on this as we build more complex projects in the Yariverse (too soon for that, girl), including our own LLM.
Rule #2: Explore, But Don’t Over Rely
Use LLMs as a tool for exploration, but don’t become reliant. This is also an important life lesson. Remember, you only have your intern for a short period of time, and LLMs are not allowed during interviews. Don’t let them become a limitation that makes you sound like everyone else. Seriously, don’t let LLMs or AI make you boring. Use that free time to get into a hobby or find a way to rule the world.
Mitigating Risks and Maximizing Benefits
To effectively use LLMs, consider the following:
- Understand the Limitations: Be aware that LLMs are imperfect and make mistakes. They are trained on vast amounts of data but don’t possess full understanding or common sense.
- Provide Clear and Specific Prompts: The more specific your prompts, the better your results. Avoid ambiguity and provide context whenever possible. It’s like how you order your coffee, be highly specific.
- Verify Information from Multiple Sources: Don’t rely solely on the output of an LLM. Double-check references and accuracy. It’s like how your bartender asks for ID before serving you a drink.
- Use LLMs for Specific Tasks: LLMs are best suited for tasks such as generating text, summarizing information, and brainstorming; but always apply your own critical thinking. They assist, not replace, your expertise.
- Continuously Evaluate and Refine: As you use LLMs, continuously evaluate their performance and refine your prompts and workflows to improve the quality of the output.
Final Thoughts
I hope that from this you learn to be responsible.
LLMs and AI are powerful tools that can enhance productivity and creativity in any project you work on.
However, always approach them with a balanced perspective and understand their limitations and risks. By following the guidelines outlined here, you can effectively leverage them while maintaining critical thinking and ensuring the accuracy of your work.
Remember to treat LLMs and AI like interns, valuable resources that require guidance, supervision, and a healthy dose of doubt.
The Yariverse 