For decades we've tidied computer files into folders, neatly organizing our digital lives so our documents and photos are close at hand.
But could the busywork of rooting through files and folders be on its way out, destined to become a relic of computing, as unfamiliar to the average user as the command line and resolving IRQ conflicts?
The killer blow to the traditional file system could be dealt by machine learning models that draw on information about what we're doing to show us the files we need at that time.
Google recently made its Quick Access feature generally available to users of its cloud storage Google Drive. Quick Access uses deep neural networks to show Drive users what it estimates are most useful documents at that time, based on information such as user activity in Drive and meetings on their Calendar.
Speaking to TechRepublic recently, Prabhakar Raghavan, VP of G-Suite for Google Cloud, pulled out his Android phone to show how Quick Access had automatically suggested the script for his keynote at the Google Cloud Next event in London that morning and the slide he would need in the afternoon.
"In the few months since we launched, we've found that 40% of all files opened in Google Drive are from here," he said of the Quick Access feature, saying it had typically reduced the time users spent looking for files from about 45 to 15 seconds.
There is also scope for the machine learning models Quick Access uses to locate files to become more accurate.
"These models are not yet deeply personalized. They are generic models that we have built for all our users. It's simply based on my usage patterns and my co-workers usage patterns," said Raghavan.
In the near future Google hopes to speed up rate at which files are presented to the user to "tens of milliseconds", and to use information related to social connections, organizational hierarchy, and topics in documents to further improve the accuracy of Quick Access.
"New signals, we hypothesize, will greatly increase the quality of the Quick Access predictions even further beyond the baseline than it is now," said Mike Procopio, senior software engineer at Google earlier this year.
Location is another piece of information that Raghavan suggests could be a useful determiner of what files to suggest, for example, knowing whether someone is at home or in the office could determine whether that person wants to open the slide deck for a meeting on their machine.
There are also always going to be more tech-savvy users who want to root around in the file system, and Google isn't suggesting replacing it wholesale, rather continuing to improve its prediction system so it becomes largely unnecessary for the typical user to delve back into files and folders.
Raghavan said: "Our goal is to be able to predict more and more, better and better. You should be completely oblivious to the fact that you have 10 million files. It shouldn't be a chore for you."
- Big data in 2017: AI, machine learning, cloud, IoT, and more (TechRepublic)
- The death of the smartphone is further away than you think. And there is no 'Next Big Thing' (ZDNet)
- AI tool for Slack gives managers insight into employees' emotions (TechRepublic)
- McKinsey: AI, jobs, and workforce automation (ZDNet)
- Video: When robots do low-level jobs, will employees really take on new roles? (TechRepublic)
Nick Heath is chief reporter for TechRepublic. He writes about the technology that IT decision makers need to know about, and the latest happenings in the European tech scene.