This could be the hack that gets rid of some of my frustration with copilot. I will have a go with this and report back. Thanks for the inspiration.

@Junior Sampson unfortunately copilot is not able to follow that instruction however I can switch into terse mode or ultra terse mode. Trying that and it’s certainly cutting out some of the waffle.

@Richard Pascual that’s a really well considered response and very much in line with my thoughts. However my testing has shown me that you can get vastly different results depending on which LLM you are using so they are not all created equal. For the privacy and data sovereignty issue nothing beats self containment - and it’s perfect for use cases where you wish to tightly integrate with your socials and even banking apps. If your AI has access to those details then keeping them secure has to be a priority. Interestingly and just based on general observation I would say that the ‘big’ hosted LLMs are spending most of their time answering relatively trivial questions so I think as AI adoption increases - cost optimisation will become a factor and that may drive further demand for local models.

All I see is opportunity- however I am the outlier in my workplace which is surprising when you consider I’m more of the oldest. I’ve adopted an AI first principle with work tasks where that becomes my go to and it’s freeing up more time for me to research and learn. That’s a win win.

Nice poll - I chose Other because my goal with AI isn’t just efficiency. It’s to build things that don’t exist yet. I’m here for innovation and fresh ideas and approaches

I started the idea in a specific context but quickly realized that it can apply to almost all generative AI output so now it’s becoming a framework in my mind. The use of LLM as a judge is also a key here so I see a continuous loop of validation and testing rather than a one shot compliance report.

Yes, I have recently joined.