Arranging scientific articles

Its 2026, AI is booming, yet there is not enough clear boundary of which articles are not written, which ideas are not explored, and which are. In the space of articles, an article is just an dot, a single high dimensional dot, for instance text-image embedding of article.

Yet these plentiful ideas are restricted to access with a search engine via google scholar. However, it does not scale with number of articles available for a human to process it within relevent time scale. From the entire global dataset of science in all of its existence, it is only there because it represetns the true reality of the universe, given the conditions (including of the author as human being as wrote that article). Anyways, just like articles link (i.e. cite) older articles, and add their own novelty, to create a new point of its own. Entire articles in existence are can be represented in the higher dimensional space perceived by human, or visible 3dimensional space, information can be shown as linked articles resulting from experiments happened from this 3d verse somewhere on this globe (or outside) at certain time in past. However, that ofc, is not the complete reality of the universe. Ass always something new can be bought into reality.

For someone who crave to be all knowing science, they indeed would need mental telepathy like access to all knowledge. The ability to retrieve information with minimal effort, and consistently verify the correct state of whats desired.

Much like the pyramid forming chain of references leading to the article, it must be visualizable so any article’s dependency is known. Currently there are certain AI tools exist, but they still list the articles in list format, and ginormous amount of scale is not seen, in which unique area they map to. An idea would be to overlay these articles in physical space for where they apply, or in application or similar topic space where they relate. Such methods are still there, tagging by keywords, AI bohrium article browser, deep search AI. However, they are not reliable to go through complete sourcing, either the AI itself must be continuously trained to retrieve all articles from its memory, as search based criteria still rely on classical search that may not give correct results.
Google Lab provides results , although they still seem to have lower confidence.

Even overleaf, when citing does not show the article being cited, its very inefficient design. When the space of articles to be cite from is in thousands or potentially more.