A field guide that also covers why we need to rethink our expectations, and what software engineering really is. A guest post by software engineer and engineering leader Addy Osmani
Fascinating read! I like how it's balancing optimism with a reality check.
I've recently explored GenAI from an unusual angle in The Executive-Coder Experiment: Returning to Code with GenAI After 7 Years. AI proved extremely useful to assist with coding. At the same time, deep software engineering background was crucial for guiding it effectively.
I regularly tell my reports that I’m not worried about the code being written. This should be the easiest part. Let’s double down on planning and design, so that we can be more informed on what we’re intentionally building rather than jumping straight to an editor and grinding away to an AI rabbit hole.
In translation, more diagrams, ERDs, API proposals, and whiteboard chats.
AI isn't making our software dramatically better because software quality was (perhaps) never primarily limited by coding speed. The hard parts of software development — understanding requirements, designing maintainable systems, handling edge cases, ensuring security and performance — still require human judgment.
What AI does do is let us iterate and experiment faster, potentially leading to better solutions through more rapid exploration. But this will only happen if we maintain our engineering discipline and use AI as a tool, not as a replacement for good software practices.
Remember: The goal isn't to write more code faster. It's to build better software. Used wisely, AI can help us do that. But it's still up to us to know what "better" means and how to get it.
So appreciate this fresh and balanced perspective on ai and using ai for software engineering, and ultimately solving our users problems / needs - which has always been the goal.
Interesting to see that we don’t yet have an answer to the junior paradox: if demand ends being only for experienced engineers, how do we create more experienced engineers? With the juniors who are no longer getting work?
AI in software development feels like a new programming abstraction layer, similar to the leap from assembly to high-level languages. It dramatically speeds up development by working at a higher level, though you occasionally need to dive into the details - just like how a C++ developer might examine assembly output for optimization. In both cases, you gain development speed while maintaining the ability to fine-tune when needed.
I would love to see a follow up episode that also explores how high level languages were first viewed in the early days and how it compares to ai coding tools today.
That is nicely said, Addy; it makes a lot of sense to be careful with AI tools, know, and know what to use and for what not. There is also this interesting view from Primeagen which is true, as LLMs are a statistical models: https://www.youtube.com/watch?v=D9q5wnX2_DY
Fascinating read! I like how it's balancing optimism with a reality check.
I've recently explored GenAI from an unusual angle in The Executive-Coder Experiment: Returning to Code with GenAI After 7 Years. AI proved extremely useful to assist with coding. At the same time, deep software engineering background was crucial for guiding it effectively.
I wrote about the full experiment here: https://www.ivankusalic.com/executive-coder-experiment/
I regularly tell my reports that I’m not worried about the code being written. This should be the easiest part. Let’s double down on planning and design, so that we can be more informed on what we’re intentionally building rather than jumping straight to an editor and grinding away to an AI rabbit hole.
In translation, more diagrams, ERDs, API proposals, and whiteboard chats.
This 1000%.
AI isn't making our software dramatically better because software quality was (perhaps) never primarily limited by coding speed. The hard parts of software development — understanding requirements, designing maintainable systems, handling edge cases, ensuring security and performance — still require human judgment.
What AI does do is let us iterate and experiment faster, potentially leading to better solutions through more rapid exploration. But this will only happen if we maintain our engineering discipline and use AI as a tool, not as a replacement for good software practices.
Remember: The goal isn't to write more code faster. It's to build better software. Used wisely, AI can help us do that. But it's still up to us to know what "better" means and how to get it.
So appreciate this fresh and balanced perspective on ai and using ai for software engineering, and ultimately solving our users problems / needs - which has always been the goal.
Thanks for the great read, Andy and Gergely!
I think this is the right take and the distinctions between how seniors and juniors use these tools is insightful.
Interesting to see that we don’t yet have an answer to the junior paradox: if demand ends being only for experienced engineers, how do we create more experienced engineers? With the juniors who are no longer getting work?
AI in software development feels like a new programming abstraction layer, similar to the leap from assembly to high-level languages. It dramatically speeds up development by working at a higher level, though you occasionally need to dive into the details - just like how a C++ developer might examine assembly output for optimization. In both cases, you gain development speed while maintaining the ability to fine-tune when needed.
I would love to see a follow up episode that also explores how high level languages were first viewed in the early days and how it compares to ai coding tools today.
That is nicely said, Addy; it makes a lot of sense to be careful with AI tools, know, and know what to use and for what not. There is also this interesting view from Primeagen which is true, as LLMs are a statistical models: https://www.youtube.com/watch?v=D9q5wnX2_DY