I just got back from Bucharest, where one of my sessions at GoTech.World was about what I believe is going to happen to developers in the age of AI.
Excitement
Excitement is really up there. I’ve been discussing AI at large with so many organisation for the past year, and there isn’t a single one of them that’s not excited about what AI can do for developers. They are at the frontline of this new technology, and the results for who embraces it properly are really out there to see.
My view of the world is very simple: with AI in their workflow developers can focus on what really matters in their day to day job, leaving repeatitive and time-consuming tasks to a Copilot of their choice. The productivity benefits are sizeable. People are happier. It’s a win-win from any angle.
When it comes to AI in new or existing systems, it’s still exciting. The potential market advantage of disrupting and accelerating existing processes in any industry is there for anyone to exploit, and it’s worth remembering organisations sit on a trove of data, which is essential fuel for a LLM. It’s up to them using this data properly.
Confusion
People are confused about the future, in particular about how development tools and platforms will evolve, potentially removing some skills and tasks from the usual day-to-day work. What’s going to happen when you have tools that today provide advice (GitHub Copilot) and tomorrow will evolve into headless Pull Requests (GitHub Copilot Workspace) for your developers to review?
Skills will evolve and adapt, undoubtedly. Over the past 12 months, for example, we introduced concepts such as responsible review and watermarking in the developer workflow. Prompt Engineering breached from the LLM world and is now essential for the GenAI-assisted developer to bring the most out of their Copilot. There is going to be a lot more permeating the developer space in the coming days and months.
Unease
Let’s be frank - the first source of developer resistance is fear of being laid-off due to the impact GenAI has on development teams. There is some risk for some types of developers, of course. But as I mentioned in the past, reducing teams by proportion of development productivity impact is a very short-sighted idea that I try to defuse whenever I can. There is simply no direct correlation as every developer is unique and operates in a different way.
Moreover, even AI needs developers, and not for training or refining. LLMs need to be harnessed and managed like any other resource, bringing DevOps and hyperautomation concepts into the mix. To name an example, a developer writing pipelines and automations for LLMOps is already AI-adjacent and they will be fundamental for any form of development of said system.
There is a lot on the horizon - patterns, products, services, etc. The world feels in a permament cliff-edge about how roles will evolve. Yet, it’s still early days to predict what’s going to be the future job description of a developer in 5 years’ time. It’s fair to say there is a mixture of excitement, confusion and, for some, even unease. However in my opinion the future is bright - there is going to be so much to do for a developer! I am looking forward to what’s next in the industry 😀