Since the rise of AI chatbots like ChatGPT and Gemini, the role of software developers has changed significantly. Instead of spending all their time writing code, many now rely on AI tools to generate it for them.
Commenting on the growing use of AI tools for coding, GitHub CEO Thomas Dohmke said that the most experienced and advanced developers have already “moved from writing code to architecting and verifying the implementation work that is carried out by AI agents.”
In his blog, Dohmke said that they spoke with 22 developers who have heavily incorporated AI tools in their workflow to understand how the technology has changed the landscape and see how things are going. Referring to these interviews, the GitHub CEO said developers were initially skeptic of AI tools and used them in small tasks and questions as they were prone to making errors.
But when large language models started getting better, developers used these very tools for things like debugging, boilerplate and creating snippets. However, with practice, they started to brainstorm with AI tools to complete more complex tasks and embraced iterative prompting.
Later on, they started collaborating with AI to learn its “thought process” and held discussions and internal demos for sharing ideal prompts and use cases. In the last stage, developers started treating AI tools as their partners for things like developing features and completing complex tasks. At this stage, many developers said they relied on AI agents to write and curate code and instead focused on refining prompts and verifying if the code works as intended.
When Dohmke asked developers about the prospect of AI writing around 90 per cent of their code, half of them said that a scenario where AI writes the majority of the code is not only feasible, but that it will happen in the next five years, while the other half said it would happen within two years.
© IE Online Media Services Pvt Ltd