AI / LLMs are a very relevant topic in any Linux Forum, I have to admit I haven’t read this entire thread. I have solved many problems over many years with internet searches, and I would have no understanding of Linux without it. The issue I have had with AI solving my problems is a lack of understanding the solution prior to it being applied. Using a AI enabled terminal can solve many problem, but a web search requires some reading. An example from my recent experience could illustrate the point here. I have an Intel processor with internal graphics (Arrow Lake “Ultra Core 9”). AI suggested a kernel parameter to force the use of the newer Xe module over the older i915 module”. This is an easy solution to find on the internet (WWW) as well. But the correct solution is to enable Internal Graphics in the firmware settings. Then any kernel will use the newer Xe module. My limited understanding of computer hardware and the Linux kernel enabled me to come up with the correct solution. The AI assist was perfectly “happy” with the solution it gave and many forums agreed. As a test I told the AI the solution I chose to use. It replied that it was actually a better solution than the one it had suggested. This actually changed my perspective of the potential of AI assisted configuration by realizing it is evolving. Linux has come a long way over the last two plus decade. I remember having to manually partition the disk prior to installation, configure X11 from the command line and create users. Now the installation process is simpler because some coder wrote a graphical installation program. When I first used a GUI to install Linux it was impressed at how easy it was. But you can click through all the defaults and have a usable system without knowing what went on behind the scene. I think of LLMs in the same way, code to make tasks easier, but they can’t replace the need for understanding what you want and need the computer to be able to do.