What is Vibe Coding? The Next Evolution in AI-Assisted Programming Techniques

You must have heard of the popular Generation Z term “Vibe”, which describes the overall feeling or atmosphere of a person, place, or situation. But what if we told you that the term has taken over the AI ​​and computing world? Do you believe it?

Ambient coding, yes, yes! This is a new concept in which you tell AI your problem in a timely manner, and then LLM writes computer code for you.

You are watching: What is Vibe Coding? The Next Evolution in AI-Assisted Programming Techniques

For example, you provide a prompt: “Write a function that takes two numbers as input and returns the sum of them”, and AI provides you with Python code (or code in other programming languages) to define the functions that are executed.

Therefore, AI is not focusing on how to translate that description into actual code. This makes the programmer’s role be to evaluate whether the code meets its requirements.

But do you know where the word comes from? Who invented it? How does it revolutionize the LLM we use? Let us know!

What is atmosphere coding?

The term “Vibe encoding” was introduced by renowned computer scientist Andrej Karpathy, co-founder of Openai and former Tesla’s AI leader. The term became well-known in February 2025.

See more : Word Search Puzzle: Find the word ”WINGS” in 7 seconds!

In short, Vibe encoding is an LLM-dependent encoding method that allows programmers to generate working code by providing natural language descriptions rather than writing manually.

He describes his method as a conversation, using voice commands when AI generates code. “It’s not really coding – I just see things, talking, running things and copying things, and mostly valid.”

Additionally, he added that Vibe encoding has its own limitations and that AIS does not always reliably fix or understand errors, so he is always trying irrelevant changes until the problem is solved.

He concluded that he found the technology “not bad for throwing away the weekend project” and described it as “funny.”

How atmosphere coding works: a step-by-step guide

Vibe encoding utilizes the capabilities of the Large Language Model (LLMS) to generate code from natural language descriptions. Here is a step-by-step guide to the process of understanding:

Question definition:

• The problem to be solved using code is clearly defined. This involves understanding the requirements, inputs and expected outputs of the software.
• Example: “Create a function that arranges a list of numbers in ascending order.”

Tip to create:

• Make a concise and clear tip that describes the problem as LLM. Tips should contain enough details to understand the AI ​​to understand the task.
• Key Element: Specifies the required functions, input parameters, and expected output.
• Example tip: “Writing a python function called sort_numbers that takes a list of integers as input list and returns a new list where integers are sorted in ascending order.”

LLM code generation:

• Enter the prompt into the encoding-tuned LLM (e.g., Codex, GPT-4). LLM processes the prompt and generates the corresponding source code.
• LLM uses its training data to predict the most likely code that satisfies the prompt.
• The generated code can include comments and instructions according to the context of the prompt.

Code Test:

Thoroughly test the generated code to ensure it meets the requirements. This involves creating test cases covering a variety of situations, including edge cases and error conditions.

See more : Observation Skills Test: Can you find the number 94 among 99 in 10 seconds?

Use a test framework (e.g. Unitest in Python) to automate the test process.

Sample test:

assert sort_numbers([3, 1, 4, 1, 5, 9, 2, 6])== [1, 1, 2, 3, 4, 5, 6, 9]

assert sort_numbers([])== []

assert sort_numbers([-1, 0, 1])== [-1, 0, 1]

Improvement and iteration:

• If the code does not execute as expected, please refine the prompts and regenerate the code. This iterative process involves analyzing errors and adjusting the prompts to guide the LLM towards the right solution.
• Modify the prompts to provide more specific instructions or constraints.
• Example: “Writing a python function called sort_numbers that takes a list of integers as input list and returns a new list with integers sorted in ascending order. The function should use the QuickSort algorithm.”

Code integration:

• Once the generated code has passed all the tests, integrate it into your project. This may involve tuning the code to fit the existing architecture and ensuring it works seamlessly with other components.
• Follow coding best practices for readability and maintainability.
• Add comments and documentation to explain the purpose and functionality of the code.

Monitoring and Maintenance:

• Continuously monitor the performance of the integrated code and resolve any issues that arise. Update code regularly to merge new features, improve efficiency and fix bugs.
• Use a version control system (such as GIT) to track changes and collaborate with other developers.
• Implement logging and error tracking to quickly identify and resolve issues.

Benefits of atmosphere coding: efficiency and innovation

Vibe encoding involves describing the problem as AI and then generating code. You can test the code, refine the description when needed, and then integrate the working code into your project. Here are its useful benefits:

1. Improve efficiency: Vibe encoding can greatly reduce development time by automating repeated coding tasks. AI-powered tools quickly generate code, allowing developers to focus on refining and optimization rather than starting from scratch. This results in faster project completion and faster iteration.
2. Enhanced accessibility: Vibe encoding reduces barriers to entry to software development, allowing it to be used by individuals with limited programming experience. Non-technical stakeholders can contribute by providing natural language instructions, thereby facilitating better collaboration between technical and non-technical teams.
3. Accelerated Prototyping: Vibe encoding can be quickly prototyping, allowing developers to quickly create and test new ideas. AI aid reduces the time and resources required to build functional prototypes, thereby facilitating faster innovation and experimentation. Startups can bring minimal viable products (MVPs) to life faster, validating concepts and effectively collecting feedback.
4. Focus on Innovation: By automating daily tasks, Vibe coding enables developers to release focus on creative issues and strategic thinking. This focus shift allows teams to explore new approaches, experiment with cutting-edge technologies, and develop innovative solutions to drive business value.
5. Improved code quality: While careful supervision is required, Vibe encoding can improve code quality by providing instant advice, highlighting errors and even writing tests. AI tools help ensure code adheres to best practices and reduces the risk of vulnerability, resulting in more robust and maintainable software.

In summary, Vibe encoding automates tasks, improves accessibility to non-technical users, accelerates prototypes, promotes innovation by freeing developers’ time, and improves code quality with proper supervision.

Source: https://dinhtienhoang.edu.vn
Category: Optical Illusion