Research trends: Reinventing the human-document relationship

April 18, 2022

Tags: Document Intelligence

Wondering which trends and ideas will shape technology in 2022 and beyond? We talked to a few members of the Adobe Research team to find out.

Tong Sun, Document Intelligence Lab Director, is tapping into artificial intelligence and machine learning to unlock the hidden value of documents. She talked to us about how documents—one of the oldest human technologies—are evolving along with the new ways we consume and share information.

Before we get into the trends you’re following, can you help us understand what document intelligence means?

Sure. You can think of a document as a container of information that you can easily share. The history of the document goes back to when humans invented language. And over time, the big thing that’s changed is the medium—lightweight paper made communication easier and more portable. The next shift was printing, which you can do at scale. Then came sharing documents over the web. Web 2.0 brought new ways to create documents—especially with social media and mobile. And more recently the cloud-based solutions are dominant. Emerging Web 3.0 will be a paradigm shift to a fully decentralized internet.

With document intelligence, we’re bringing in artificial intelligence (AI) and machine learning (ML) for the next step of document evolution. This will transform the human-document relationship, including how we create, share, and collaborate with documents.

What trends are you following in document intelligence right now?

We’re in an age of abundant data because of the web, and compute infrastructure has become so efficient. These things are powering deep learning with AI and ML, including computer vision and language technology. But computer vision has been primarily focused on things people see in the natural world. So far, we don’t have an equivalent ImageNet dataset for documents, especially for business documents in the enterprise world.

So the next trend I’m following is how we can apply AI and ML to deeply understand documents. It’s important because there is a lot of “intelligence” locked in documents—it’s very hard to search for things once you create a document. We want to liberate that content so people can search and interact with it. This is document understanding.

Once we liberate the content, the next question is how to design and enable novel experiences. Right now, we open a document and read. We also have audiobooks, or we can listen to a lecture or find a video on YouTube. But what about when we want to interact with a particular document while we’re moving around? I might be reading a book in the office, but I then I’m on the go and I want to listen. So the next trend I’m watching is how we’ll create experiences that are based on the user’s context.

What kinds of research questions do you think will define document intelligence over the next 10 years?

The COVID-19 pandemic has accelerated digital transformations in all walks of our life, including how we interact with others, work, shop, and conduct daily activities. So reinventing how humans interact with documents in authoring, consumption, and collaboration is critical. With this in mind, research at Adobe is exploring dynamic and interactive documents in AR/VR environments for immersive reading experiences and cross-modal collaborations.

Adobe previewed Project Dually Noted at Adobe Summit Sneaks 2020.

Another way to think about the future is to consider why documents have endured this long. Documents have been with us for so long because they’re a format for human communication, but what about the future, when your colleague could be a robot? What forms of communication will we need? Will it be human language? Or computer language? Should we send commands to a robot in a file or have a natural conversation with a robot? To take it one step further, will robots perceive human emotions, understand intent, and intelligently collaborate with humans?

There are lots of other forces that will influence the future of the document, especially the future of the internet. I’m interested in the new ways we’ll use the portable nature of documents to create, communicate, and share our stories.

Which trends are you following outside of the document intelligence world?

I’m interested in the future of the internet, especially the decentralized nature of Web 3.0. I’m thinking about digital identity and things like blockchain—not just blockchain itself, but the movement behind it, including new forms of digital identity and new ways of payment like cryptocurrency which are all based on the notion of decentralization.

I’m also interested in what will happen with the metaverse. Back in 2008, we thought Second Life—the first massive 3D virtual world—was the new paradigm. But it was too early. It wasn’t very immersive and it was very slow. So I wonder, is this the right time? Will the metaverse become a real thing? I’m still skeptical more than 10 years later, but I see how the technology has evolved and I have some faith in it now. It’s all quite interesting to me.

Wondering what else is going on in the world of document intelligence research at Adobe Research? You can learn more here.

Related Posts