Jonathan De Vita graduated from Lancaster University with a BA in computer science, having studied AI and coding using specialist programming languages. This article will examine the potential impact of AI advancements on the field of computer coding and how AI is revolutionising the practice of code writing.
Like many other industries today, AI is transforming the world of software development, fuelling speculation that code writing may become fully automated before very long. Rather than being purely intuitive, like writing poetry, writing code is a highly technical process. Experts suggest that AI coding could democratise the process of creating software, enabling less tech-savvy individuals to leverage AI to produce certain kinds of code for them.
In an interview published by Forbes in January 2024, a group of MIT undergraduate students was asked to predict whether AI would be writing the majority of the world’s code by the end of the year. While 30% of respondents agreed that 3–5 years was a more likely time frame for the majority of code to be written by AI entities, one student pointed out that it was already happening, admitting that he relied heavily on GitHub Copilot to write code for his web app. Another respondent suggested that, within five years, AI would be generating the majority of generic and functional code, estimating that 80-90% of regular application code would be generated by AI by that point. He went on to predict that the remaining 10–20% of code would be written by humans, specifically that for legacy technologies and systems and complex integrations that are not developed with standard programming.
Experts predict that while AI will be used to write the majority of coding, certain aspects will remain within the remit of human computer programmers. Pundits suggest that security in particular would not be exposed to AI. However, for other aspects, AI presents scope for significant innovation, just as the rise of the computer changed the world. In the Forbes report, the MIT team suggested that while there will be big efficiencies with AI taking the driver’s seat and creating code, there are also some potential pitfalls and challenges as technology wrestles with more creative or intuitive aspects of programming that have traditionally been the province of the human mind.
Leveraging algorithms that are trained using existing source code, AI code is not always perfect but can provide a useful starting point for developers. Tech companies like OpenAI are scrambling to develop a new generation of AI assistants capable of not only writing code but optimising, debugging and critiquing it too. By providing a suitable starting point for human program developers, AI code allows them to develop software more quickly and efficiently than ever before.
In software development, AI code generation can be utilised to optimise the process by providing autocomplete predictions for repetitive and boilerplate coding patterns. By providing these suggestions, AI could save human developers considerable time, effort and resources, eliminating the need to trawl the internet for the requisite information.
Using natural language processing to generate code alongside AI capabilities to detect potential bugs, AI could be leveraged by software developers to help them analyse code, identify potential issues and suggest further code tests. Ultimately, AI code could enable human developers to bring software to market more quickly, as well as paving the way for significant cost savings.
In addition to code generation, AI also helps increase productivity and speed throughout the software development process, automating mundane and repetitive tasks. Providing real-time feedback on customer interactions, AI minimises the potential for human error and biases in decision-making processes, enabling developers to test user scenarios at scale.
