The Opportunities (and Concerns) of Vibe Coding by Dr. Timothy Smith

Photo Source: Unsplash

If you have not heard of it yet, you will—vibe coding. Vibe coding has come of age, with the Collins Dictionary selecting it as the word of the year for 2025. Vibe coding refers to a new type of computer coding, supported by large language models, that shifts the way computer programs are written. Traditionally, computer programmers begin by learning the fundamentals of programming, including safe and efficient coding practices, and by developing fluency in various programming languages. After learning to write computer code, programmers will tackle problems by first writing out how they will solve them in code, a technique called pseudocode. From the pseudocode, the programmer will build their computer code to solve the initial problem, drawing on their training and expertise. On the other hand, vibe coding involves the user describing, in natural language, the situation she needs to solve to a large language model chatbot, which then maps the problem and generates computer code to solve it. The vibe coder will iterate with the chatbot to refine the code until a satisfactory solution arises.

Vibe coding effectively puts the complex, at times arcane, world of computer programming into the hands of people who may have never even taken a computer programming class. Many people know their problems very well but do not consider using computer programming, machine learning, and data science as potential solutions. With vibe coding, non-programmers have a robust set of new tools at their fingertips. For example, I recently wanted to explore Shor’s Algorithm. Shor’s Algorithm helps efficiently find the prime factors of a number, but I did not have the time to write a computer program from scratch. Fortunately, with some vibe coding, I asked Copilot to generate code to allow me to input numbers to initiate Shor’s Algorithm and print out the results of the calculation using the computer language Python. In less than ten minutes, I had a functioning program! Vibe coding can also help in many other areas, such as game design, analyzing data in an Excel sheet, organizing information, and much more.

As with all technological advances, new risks accompany the benefits. Vibe coding often entices individuals without a computer science background to try coding. We have seen the benefits and opportunities available to the novice coder. Still, vibe coding can also lead to serious security and performance issues that may not occur to the inexperienced vibe coder. For example, a vibe coder might create an app that pulls back patient health information from a secure database if the patient provides the correct security information. Still, the vibe coder may not know how to ask the chatbot to make a secure query interface. Unfortunately, hackers who detect an insecure interface can steal data or even damage or delete valuable data from the database. In another case, a vibe coder may work with a chatbot to automate their stock trading based on what the vibe coder feels to be the proper parameters; however, when the vibe code interacts directly with the market, unforeseen conditions may arise that cause the program to buy thousands of shares at a significant loss.

Vibe coding represents a new frontier for non-computer scientists to leverage the power of computing without any formal training. Under the right conditions, vibe coding opens up tremendous possibilities previously walled off from the amateur coder by a lack of training and experience. Vibe coding works best with strong intuition for solving low-stakes problems and an opportunity to iterate on the solution without harm. On the other hand, vibe coding should ultimately involve an expert in high-stakes situations such as financial, health, or security cases, where unforeseen bugs in the program can lead to irreversible, unrecoverable harm. I encourage everyone to have fun with vibe coding, keeping in mind when to call in the experts.

Dr. Smith’s career in scientific and information research spans the areas of bioinformatics, artificial intelligence, toxicology, and chemistry. He has published a number of peer-reviewed scientific papers. He has worked over the past seventeen years developing advanced analytics, machine learning, and knowledge management tools to enable research and support high-level decision making. Tim completed his Ph.D. in Toxicology at Cornell University and a Bachelor of Science in chemistry from the University of Washington.

You can buy his book on Amazon in paperback and in kindle format here.