Within the last few months it has been ChatGPT everywhere.

We are always eager to explore new tools and technologies that can enhance our workflow and deliver better results to our clients. That said, the number of Instagram advertisements from people claiming that ChatGPT can help me make $100,000 in 60 minutes definitely made me approach trying out the new tool with a healthy dose of skepticism.

Here are our initial experiences with ChatGPT and what I can see as its current implications for code development. Since the ChatGPT is always learning, changing and improving, at 14 Oranges, we’re viewing our exploration of using it for development and coding purposes as an ongoing journey. We are keeping a close eye on its evolution, so we can maximize any potential value it can bring to our work.

Exploring the AI Potential

As developers, we encounter repetitive and time-consuming tasks during the development process. I think this is where ChatGPT has shown its potential to be a game-changer. By leveraging its capabilities, we can save significant time on mundane tasks such as setting up foundational classes and project configurations. Rather than spending hours manually coding these elements, we can use ChatGPT to generate the necessary code snippets in a matter of minutes. It’s like laying a strong foundation for a house before the actual construction begins.

The Need for Human Expertise

However, it is crucial to emphasize that while ChatGPT can assist in generating code, human expertise remains irreplaceable. Just like planning a trip, where a wrong calculation could have dire consequences, coding also requires careful analysis and understanding of the situation when the human is reviewing its results. ChatGPT is not infallible and may produce code that is not optimal or prone to issues like SEO penalties or security vulnerabilities. It is imperative that experienced developers review and validate the generated code to ensure its quality.

Learning and Testing

Our initial experiences with ChatGPT have been both enlightening and cautionary. We have witnessed instances where inexperienced users blindly implement code snippets from platforms like Stack Overflow, without comprehending their underlying implications. There are often many nuances in code that only a trained developer might notice that could have implications that an inexperienced developer or non-developer will not be able to troubleshoot.

For example, you could ask ChatGPT to write the code for a WordPress plugin. It would write some code and you use it and it seems to work. But, without fully understanding the nuance of the code, that new plugin could be having implications that you don’t even know about – negatively impacting your SEO or causing a security vulnerability, or conflicting with another plugin, which may alter the ability for your forms to work correctly on your website, for example.

At this point, it’s still important to maintain a balance between the convenience (and brilliance) of ChatGPT and the need for human discernment. As the technology evolves, the frequency of code generation errors should decrease, but, especially during this transitional period, we definitely won’t be relying on ChatGPT to create any code for us for anything critical or without careful review by somebody on the 14 Oranges team.

Comparing ChatGPT to Compilers

When I consider the potential trajectory of ChatGPT, I find it helpful to draw parallels with the history of compilers. Decades ago, programmers had to write code in low-level assembly language, which was laborious and required extensive knowledge. However, the introduction of higher-level languages and advanced compilers transformed the programming landscape. Developers now focus on problem-solving, relying on compilers to translate their logic into machine-readable code. Similarly, we envision a future where ChatGPT assists developers in tackling complex challenges, while the actual coding process becomes less cumbersome.

Looking Ahead

While we are excited about the possibilities ChatGPT offers, we remain cautious about its current limitations. The technology is not yet foolproof and demands careful oversight. We must continue to exercise human judgment and expertise to ensure code quality and address potential pitfalls. As ChatGPT evolves and matures, we anticipate that its success rate in generating high-quality code will improve significantly.

As the machine gets better and better at writing code, over time I can imagine it getting close to 100% accuracy. At that point, it’s possible that the work of a developer becomes less about writing code and more about framing problems well and then asking ChatGPT to help you solve those problems.

At 14 Oranges, we always want to stay at the forefront of technological advancements in application development, so it was inevitable that we would explore the potential of ChatGPT. The tool has shown it can help accelerate certain aspects of code development, allowing us to save valuable time on repetitive tasks. But we’re still in the very early days of learning how best it can be used.

For the foreseeable future, we emphasize the need for human expertise to review and validate the generated code, as ChatGPT is not flawless. As we move forward, we remain excited about the prospects of ChatGPT and its role in revolutionizing the development process. We will continue to monitor its progress and integrate it into our workflow responsibly, ensuring the delivery of top-quality solutions to our clients.

If you’re looking to have an application or a website developed and aren’t a developer, I wouldn’t suggest allowing ChatGPT to take on this task for you just yet. The team at 14 Oranges would be happy to meet with you to learn what you need and share how we can help. Book a call today if you’d like to learn more.

Sylvain Marcotte is CEO and President of 14 Oranges.

The Benefits and Limitations of ChatGPT for Application Development