For context I have been using GLM (deep thinking) as a baseline to test AI and it’s usefulness.
My frustrations stem from the point that if i ask questions directly the AI has no context of what I am talking about and i need to manually direct it to the page. That’s the first thing. If i just ask questions and then want it to summarize the output so that i can save in markdown I have to copy over and reedit the entire thing in the card i want to add the details to even if i specify the output to be markdown. There is no live credit consumption counter and it is very difficult to know when i have reached the usage limit or when the task is too demanding and i should be more specific.
More often than not asking it to gather simple information about a table and mechanisms relating to that table in the document in further detail is a total headache. It just processes as much as it can and then returns with the document is too large and i can’t search (WHILE CONSUMING CREDITS). There is no estimate for how many tokens it will use to accomplish said task. So a simple task like get me further details of the mechanisms outlined in this table from the document which is a very simple use case becomes a task of trials and errors. It asks me to be more specific about the chapters i am looking for. If i knew where to find these things I wouldn’t be using AI in the first place right? Apart from asking questions directly and it pointing to unverified information from online resources or it’s own knowledge database you just run around in circles. Where you might understand the whole thing and then not know where you studied it from or whether the information is even verified as is the problem with all big search engines. They summarize everything which is useful but when deep learning i just need it to point to the information I want to study it for myself. I have tried setting up different AI templates and customizing them but to no avail. Currently I just burn through credits with no useful informational output. The guide and quiz modes arent actually useful either if it’s just gonna highlight the right answer for me.
So I want to know whether my usage of AI is unjustified or I am using the wrong settings for my study. How should I proceed. Or is it better to just stop using AI slop altogether because you can’t really allow it to add information to the mindmaps either as it just destroys the whole structure if not prompted with extreme precision thus actually being a hurdle to study rather than an aid.
Any feedback on this would be appreciated. I just want to save time by not having to gather materials before actually being able to study.