How are you navigating and adapting to Generative AI applied to Software Development?

My company uses GenAI in many of our core product offerings, and we're still very cautious about it overall.  

A key concept that seem to help here and in other places:

You are responsible and accountable for the code you submit, ensuring they have reviewed the code, it works, has appropriate test coverage, and it meets the requirements specified are all still expected.  Do not blindly accept AI-generated code.  

We've also explained that it would be a very bad look for the first time a developer reviewed their own code during a production issue where we find out this was a blind copy-paste or inclusion from Copilot or similar tools.

All that said, we're trying to encourage everyone to become familiar and proficient with many of the tools and systems.   They can provide help, and there's no need to fear them.

I remember seeing a similar sentiment when IDE's started including good context-aware suggestions (IntelliSense, etc.).  Most developers now wouldn't be able to code effectively without that level of tooling assistance.  I think these AI-based systems will find their way into workflows from Junior to Senior levels.

...

I'd also put more caution on their use in a highly regulated setting, like defence, medical device, finance, etc...  everyone should understand this is a risk/reward discussion.

We've begun to train a group of developers on tools (Github Copilot, Cursor.so) and are working with clients to implement policies and procedures to integrate generative AI as part of our core software development services.