Do you think that using AI coding tools can cause developers to feel less engaged with their work? What are some signs that this could be happening for a team member?

135 viewscircle icon1 Upvotecircle icon3 Comments
Sort by:
VP of Engineering in Manufacturing7 hours ago

I think it is important to understand that the proper use of Artificial Intelligence coding tools is likely to be a completely different type of job than what we traditionally associate with SW development. As a consequence how we engage the AI, the type of people that use the AI and the expectations for SW engineers will be different in this future world.

https://sajalsharma.com/posts/effective-ai-coding/?utm_source=tldrnewsletter

This is a fantastic article that establishes a set of best practices and makes it easy to envision what the day in the life of someone effectively using AI for coding would do. You can clearly see that this is not the behaviors of a junior engineer. Which begs the question where do all the senior engineers come from if we no longer have junior engineers.

To summarize some of the key points:
1) Interacting with the AI as an architect.
2) Storing AI prompts and documents to be consumed by the AI in the SCM repository. This allows for you to refer to documents and concepts consistently from say both sides of an interface.
3) Always challenge the AI when doing pair programming. Often it will take short cuts. Just because it is prolific doesn't mean it is right.
4) Plan first -- Allow the AI to help drive the plan/strategy for what the code/architecture will evolve to.

Personally I think were going to get back to a model where we end up with a "master AI developer" and an "apprentice AI developer". Expecting to hire people out of college capable of doing this kind of work is unrealistic. Training is required to become an expert in both the domain and the architecture. And those traits are necessary to use an AI effectively in driving productive change.

Mike

VP of Supply Chain in Finance (non-banking)7 hours ago

While I don't think you can apply any generalization to this (individuals may differ), I have generally seen Developers feel quite excited about using AI coding tools. There can be a lot of coding that is very mundane, and AI can eliminate that burden allowing them to get after more challenging tasks. AI tools (similar to auto-complete capabilities) often suggest ways that developers hadn't considered which allows them to think through the best solutions. (Let's face it - many of them were just looking up ways to solve issues using StackOverflow anyway).
But given there is a reason you are asking the question - what you might be experiencing is something where the developer is feeling replaced. What this may mean is that it's time for them to shift from just being a developer, to adopting more of an Software Engineer mindset. While this may seem subtle, I have used it when talking with my teams that developers often just ask for someone (or some user story) to tell them everything they need and then they go code it. (AI can do that now). Software Engineers, on the other hand, often consume to problem, think through the various solutions, think about the building blocks that are needed as a part of the architecture, but don't focus as much on the coding time itself. They are problem solvers, code is just a means to solve the problem, and many tools have been assisting with writing the code for a long time. Look for those who are not engaging in understanding how to solve problems, understand customer impacts, or thinking through different solution patterns, etc. Those might be people who feel like they've just been replaced by technology.

VP of Sales7 hours ago

No, in a prior role. I was heavily involved with the Dev team, and the CIO pushed them to use AI for coding and then to clean it up, as AI has a tendency to overbuild—too much unnecessary coding. The one thing I have learned is that you need a strong SRUM lead and daily cadence to hit timelines and keep the team engaged and on track.

Content you might like

Lack of visibility16%

No data oversight39%

Unknown code and security controls34%

Risk of data exposure7%

Other2%

View Results

Yes.54%

Haven't decided yet.33%

No.12%

View Results