Where should our digital data go after we die? UBC study explores possibilities

Science, Health & Technology

People want control over what personal digital data is passed along after they die, along with tools to make it easier to do so, according to a new case study by computer science researchers at the University of British Columbia.

On the other hand, people found the idea of creating AI-powered replicas of a deceased person “creepy.”

The study’s full findings — among the first to look at ways of preparing personal digital data for death – will be presented next week at the 2021 Human Computer Interaction Conference (CHI.)

“As someone growing up completely in the digital revolution, all the memories captured of my life are stored digitally. It struck me that many of the platforms I use don’t have great tools to support that data after I’m gone,” said lead author Janet Chen (she/her), who was an undergraduate student in the department of computer science at the time of the study. “We wanted to look at how to curate this data both while living, and after death.”

Using a method called ‘research through design,’ Chen and her co-authors, Francesco Vitale and Dr. Joanna McGrenere, created 12 rough design concepts for data management. The concepts were then presented to study participants, aged 18 to 81.

The researchers also explored different levels of user-control by presenting human-selected, computer-selected and AI-powered alternatives. These employed techniques such as nudging the user to complete tasks, collaborating with family and friends and gamifying the process.

Some of the concepts leveraged design elements from existing tools, such as videos that are auto-generated from iPhone photos. One entirely new concept was ‘Generation Cloud,’ whereby a family could upload meaningful data to something like Google Drive, then access it, or even contribute more, in the future.

Other tools included:

• Checklist Crusher: An interactive tracker to tackle data, with target areas such as bank statements and certificates suggested by the system. Over time, users gained enough points to share trophies and unlock free storage space in the cloud.
• Future stories: A way to post data in the future by allowing users to upload photos, messages, or songs — and then pick a date at least 50 years into the future. The contacts chosen would be able to see the story on that date.
• Memory Swipe: A game in which user-selected photos, recipes, games and certificates show on a screen as the audience votes on which ones to keep.

[View the workbook with all 12 conceptual tools here]

“Users were receptive to the autogenerated video tool, because they could easily pass a pre-made video along to family and friends, as well as really liking Generation Cloud,” said Chen. “One concept people clearly did not like at all was an AI-powered replica of the deceased person, which would interact with future generations. They said it was scary and creepy.”

Overall, the researchers found that participants had not really thought about their digital data, but when presented with concepts, preferred ideas that allowed them to preserve their sense of agency and control over what was passed along, with tools to make that process easier.

At the moment, there are only a few tools available that consider data after death, such as Facebook’s Legacy Contact or Google’s Inactive Account manager, but the researchers say these are platform-specific and limited in range.

“Ten years from now it will likely be quite commonplace for people to be thinking about the reams of data they have online, and there is a huge opportunity for research with larger populations and new interfaces to support people who care about what happens to their data,” said Dr. McGrenere (she/her), a professor in the department of computer science. “Tools need to be lightweight to use, and designed so that they support the range of individual differences that we saw in our participants in terms of how they want to manage their data.”

The paper “What Happens After Death? Using a Design Workbook to Understand User Expectations for Preparing their Data,” will be presented on May 11 and 12 at the 2021 Human Computer Interaction Conference (CHI.)

Interview language(s): English