John Duprey 
Thomson Reuters decided to adopt AI-assisted development through the use of GitHub Copilot (and other tools like AWS Q). As the adoption champion for Thomson Reuters Labs, I helped get our engineers onboarded. Overall, the tool was well-received. Even those who didn’t initially feel the need for it generally agreed it was useful for removing boilerplate coding. However, some complained that it occasionally took more time to fix generated code than to write it directly.
I recently published a blog post on the subject: “AI-Assisted Development: Redefining How (and When) We Code”
In this blog, I discuss the current and emerging capabilities of AI-assisted development tools, where they currently excel, and where they tend to falter. I also explore the pitfalls of using these tools and their implications for the software development profession.
I routinely hear questions about whether LLMs signal the end of the software engineering profession. I don’t think they do.
Coding assistants won’t replace most software engineers, but they will require adaptation. Engineers will shift from routine tasks to areas where human intelligence excels — learning, adapting, and creatively solving problems with code.
Whether AI will someday replace the need for all coding is still TBD. But I don’t think that will happen any time soon.