The proposed filter-website is aimed to reduce the amount of AI hallucinations to the minimum or fully eliminate, is created to be integrated with AI models and future AGI.
To be honest, I had this project before, but did not know where to use it. So let me talk about my inspiration. It was 2 years ago, when I tried to solve Liars paradox. I was trying and trying, but no ideas, no results. Then suddenly I came up with contextual operator, after which I started developing my system.
The system I proposed is aimed to eliminate the amount of AI hallucinations. It is integrated with AI. So how it works: an user asks a requests from AI, these implemented filter checks it whether the information is true, false or undefined. If it is true, it confirms. If it is false, it corrects the user. If it is undefined, he directly claims that it is unknown, but if the user wants to use paradoxical statements, the filter automatically blocks it by sending it into U set.
As I mentioned before, the idea started from contextual operator. So, after, I discovered its types for each set (Undefined, True, False). If K=K+, then S€T (true set), if K=K-, then S€F (false). However if K=K_0, then S€U (Undefined). Also each operator has its boundary: K+€+Z, K-€-Z, K_0=0. Moreover, unlike other systems (Kripke, Tarskiy and so on) mine one is dynamic. The formula below explains it: K(t_1)+Delta_K=K(t_2).
The only challenge we faced was how to make our system dynamic to make one universal.
I don't know but I'm really proud of the website I established, since it really works and can be used into practise
What I learned is how to use AI tools to create a product within limited amount of time.
Actually I was thinking about it many times, and I decided to direct my focus on my other projects that I left in the foundation level.
