Do you see signs that entry-level software professionals are too dependent on GenAI — and if so, is that a problem for the future of the industry?
Sort By:
Oldest
Director of Customer Success in Hardware4 months ago
Simple answer: Highly unlikely except for specific use cases.Detailed response: Based on what I have seen and heard from my friends/colleagues, GenAI produced software is still very unreliable (50% correct at best) and requires human validation. It also does not write large pieces of code but one could get around this limitation by segmenting their code. But it is worth noting that sometimes, code validation may take longer than writing the code in the first place. There is also the risk of your code ending up in the public domain so I believe that majority of professional SW engineers will shy away from it just for this reason alone.
Where I have seen evidence of GenAI being efficient is “commenting the code” after it is written by an engineer. This is a task that most engineers dislike to do but is a good coding practice (and requirement). Again, the risk of your code ending in public domain is present.
Chief Technology Officer in Software4 months ago
I don't see it as a problem but an opportunity to grow together. Earlier entry level folks used to get knowledge transfer from senior folks hence the productivity of senior folks would be lesser whenever a new team member joins.Now they use GenAI and can write code with its help and even improve. There will always be 2 sections : 1 who learn and improvise and others who stick to GenAI to do all the work. It depends on individual to choose a bucket.
Director of Product Engineering in Software4 months ago
I don't see this as a problem right now. Associate Director of Engineering in Finance (non-banking)4 months ago
I don't see it as a problem. They will eventually realize that they will have to learn from other sources too esp. Seniors and peers.
The senior professionals who have 10+ years of experience as well as those who don't intimately know the Software Delivery Lifecycle tend to point out this as a problem, but I don't think that is the case for a few reasons:
1. It's not a new problem. Accessing previous knowledge and shooting in the dark with "copy/paste" techniques always has been something entry level professionals were doing.
2. With the proper training GenAI can be used for solution validation -> using the right prompting technique professionals can challenge the output and use GenAi for higher quality solutions
3. There are so many quality gateways in the process that is impossible to generate deep issues: code review, Unit Testing, QA, UAT, etc etc.
If anything, I think entry-level professionals should be trained as GenAi first professionals and educated to rely on GenAI tools (like assistants) to brainstorm, develop and challenge the solutions. It's going to be a new way of delivering software where syntax accuracy and knowing the libraries won't matter that much.