Tuesday, August 06, 2019

The Rise of AI in Finding Your Documents


Google just released yet another service with Artificial Intelligence (AI) running the show: Priority folders in Drive. Google has always been about searching for content; rather than, organizing content via hierarchical folder structures. Google is now adding AI to the process. Google’s back-end AI systems try to determine what it calculates to be the content you either would search for or should be aware of. In other words, it wants to suggest the content it believes you would most want. Over time, it will learn from your actions and will hone the process.

Priority is very close to the Quick View folder you would have seen on the mobile version of Google Drive. The Priority folder is really more of a view. You cannot move or create content to the Priority area. This is purely an area for Google to suggest content it estimates to be the most likely to be of use to you. Some of the elements it appears to consider when deciding what to present as priorities are
  • Recent work
  • Work that is done one a regular basis, such as a quarterly report.
  • A document that has been edited by someone working on the same document.
  • A document that has a comment.
  • A survey form in which a new response has been recorded.
Google will provide a brief explanation of why the document has been selected and the action it predicts you would want to take. Over time, Google’s AI systems should refine their selection criteria based on the actions you take – or don’t take.

Search: The New Information Paradigm

The fundamental purpose of any process of structuring content is not to organize it; rather, the purpose is to make it find-able. Many times with technology, the process is confused with the goal. The reason for keeping information is obvious: you cannot find what you don’t have. The reason for folders is to limit the search to the point that the amount of returned data is manageable. This is probably why paper file folders aren’t a foot thick or so narrow as to only accommodate a few pieces of paper. Having a file folder for each piece of paper or putting every paper into one huge folder defeats the organizational benefits of having a folder system.
The primary reason we used folders with paper documents is because it was not possible to search the text of the documents. Folders put content into manageable collections that a person could search by scanning the documents with their eyes. The move to digital format has completely changed the logic of finding information. Computers can search every document just as easily as they can search a folder. We would never consider reading every paper document to find the one we need, but computers can easily do so. Our physical limitations have effectively determined the structure of how we organized paper documents. Collections of documents could not be too granular (e.g., yesterday’s letter to joe) or not too broad (e.g., all letters from last year). Paradoxically, the result of too many folders and too few is the same. Computers allow us to redefine the organization of content. It’s important that we use the capabilities of computers to their best effect and not simply transfer our human-based filing cabinet organizational structures to the digital environment.

Because computers can scan millions of documents just as easily as they can a dozen, the need to organize content into manageable folders is no longer the same. This does not, however, mean that some sort of structure isn’t helpful. Structure, in the digital age, means both classification -- the equivalent of a paper folder system -- and process. How we organize the process by which information flows through your systems is another form of structure.

AI is extending the search paradigm by learning what users are doing and predicting what they either will search for or would have wanted to search for, if they only knew. In Star Trek, the crew was always asking the computer questions. Siri's great grandchild might do that for us, but I think it's also reasonable to expect computers to bring to our attention those things it "believes" we need.

No comments: