This past couple of weeks has been very much like Alice's descent into the rabbit hole. Unfortunately the end point wasn't Wonderland, but frustration.
My day job corporation has made a commitment to educating everyone about AI with ten weeks of courses, videos, and projects. On the very positive side, this was a kick I needed to learn more about a technology that will completely change our lives, if it hasn't already .
Our IT Team did an extensive evaluation of the various options and settled on Claude from Anthropic. When you start up, it's not quite a blank screen but it's close. I started with some basic questions and requests just to get a feel for what was needed to get a sensible response. I quickly learned a few things about giving Claude a request; be detailed and specific, and tell Claude to ask questions if it doesn't understand.
The quality of the responses as I've got better at prompts has been quite scary. I've also learned to challenge the results. As an example, I had a prompt conversation with Claude about analyzing some business documents and exporting the results to markdown so I can add them to Obsidian. Claude did a good job on that then talked about structuring the response for an obsolete reporting solution within Obsidian. What about Obsidian Bases, I asked. Claude thought for a moment, then agreed that was a better solution, and adjusted the output accordingly.
It took a couple of hours for Claude to understand and implement what I wanted. A promising start but I feel I've barely scratched the surface of what's possible.