So I was sitting there thinking about how there are all these issues with AI. I was thinking about how in the past the way I would describe AI as like dealing with an ||Extremely Smart and Charming Autistic Content Creator with Unreliable Encyclopedic Knowledge and Math Skills who's on a constant Caffeine Overdose|| (Skip Reading). But I've updated it to be simpler. It's like working with AI is exactly like working with a Kindergartner who happens to have read the entire internet or you can just say Kindergartner.
There are fixes for all of these problems. There are specific ways users are supposed to be using AI in order to get the most use out of it. The best way to do that is to set up a very easy and simple UI that handholds and leads users into the proper usage of AI.
The UI should setup the user in a Project, and have simple drag and drop options and simple yes/no frameworks that allow for deep customization through the simple yes/no or few standard. Because that is the first main fix for context length. Setting up reference documents that the chat can draw from. Gemini has managed to do that, but is still terrible with their context length in combination with their thinking power since it still fails at simple tasks. As a conversation moves along, the chat should have notes periodically saved to the project folder so, as before, the AI has easy reference points.
The biggest issue with AI right now, and is pretty much the main fix needed, is its recall power and capabilities. Before it was tool use. You seem to have fixed that, but the context limit is still effecting that negatively as well. And to identify the issue, the context limit is causing thinking issues with the AI, and the fix I hope is indexed data or reference points; Some kind of folder or filing cabinet hierarchical organization system.
I want you to hand hold users. Limit their path but don't limit their creativity. Just make it super simple and easy for them. When the user needs something done, explain to the user the simplest and easiest steps needed to be done to use the AI the best way possible. If this needs to be done then that needs to be done. If the user wants this thing, then tell the user these things need to be accomplished. Setup the workspace completely and continuously for the user.
Handhold the user throughout the entire process and never ever let the user experience any of the problems related to AI. I mean you have experts and users who have enough experience with AI to write up guides for all the fixes to all of the problems and issues with working with AI, so just implement them now.
#I figured out a big key in the marketing issue.
undone sundial