I have gone from creating a few rules in Rule Machine to directing Co-Pilot (split screen) to create a totally over-spec'd app which controls my fan against device humidity, openweather humidity, time of day. Customised triggers and polling, auto off, conditions for on. It probably has some logic errors and useless processes, but that's not the point. This has all been achieved with absolutely no programming knowledge. Yes, I cursed Co-Pilot numerous times and almost gave up a couple of times. But explain what you want and co-pilot will follow your instructions and build it. And I've learned a lot quickly.
1 Like
I find Copilot is great for working with coding projects. I use Graylog for logging from my firewall and servers and lets say that Graylogs documentation on pipelines is really lacking. But with Copilot I was able to consolidate a bunch of limited extractors into some complex refactoring of the log stream into a usable form for dashboarding and human readability. Welcome to the Vibe programming world. 
1 Like
Co-Pilot <=> Stack Overflow ??? Just easier... 
1 Like
I was literally discussing things like this with my colleagues today who are using AI for similar use-cases. Our view was if AI can deliver us / our end-users with 60-70% of what is needed, is can still significantly reduce time spent doing the simple tasks that can often take time. I think the caveat is that in our case the users of the output have enough understanding to distinguish the good from the bad. I think if you approach your use of the tool (AI) with the understanding of where your expertise is, then you can make best use of it, questioning what you need to and taking inspiration from suggestions you had not considered. If you are stepping into a new field then it is still worth cross-referencing concepts that you encounter through AI to make sure they align with what others would also recommend.
I think that is suitably vague.... 
