Artificial Intelligence and the Information Lifecycle 

Alexander Supertramp/

Records managers need to manage information in new ways. 

The year is 1989 and we’re introduced to the World Wide Web. The Berlin Wall is coming down. The Exxon Valdez is spilling oil in Prince William Sound, Alaska. Students are calling for democracy and free speech in Tiananmen Square. Crockett and Tubbs are clearing the mean streets of Miami. A future pop star by the name of Taylor Swift is born. This all occurred 30 years ago, around the same time as—if not more recently than—a number of government systems were put into place

Fast forward to 2019 and consider all the disruption that emerging technology is presenting to the federal government. Blockchain, quantum computing, internet of things, robotics, 5G … the list goes on. What does this mean? When you consider the capabilities of these new technologies and 30, 40 or 50-year-old legacy systems, agencies are generating large volumes of records, information and data in multiple formats—physical and digital—that must be leveraged and stored effectively.

No matter the format, all this information is part of a lifecycle: Agencies create it, use it, store it and destroy it. Besides the sheer volume of information, this lifecycle process is no different now than 30 years ago. The question is, how can agencies better manage that lifecycle? And, what can they put into place to glean insights from the information wherever it is within that lifecycle?

Records Managers Can Help

In light of all this information—and pressing National Archives and Records Administration electronic records deadlines—government records managers can help by:

1. Transforming to a New Way of Working

Records managers need to manage information in new ways. Many agencies today struggle with the efficiency of their records and information management (RIM) programs, requiring an investment in capital and resources. Agencies should move to a more optimized IT environment consisting of colocation and cloud services; automated business processes; and outsourcing of non-core processes. This will allow agencies to repurpose their space, reallocate their resources, and achieve a new level of digital maturity.

2. Mitigating Risk

The increase in the volume and variety of information that agencies are experiencing also exposes them to additional risk either from breach, cyberattack or loss. Agencies must mitigate this risk by not only securing where that information resides and how it’s accessed but also setting and enforcing retention policies, enabling them to know what and when they can defensibly destroy. This also better prepares them for audits or other compliance activities because they know what they have, where it resides and how long they must keep it.

3. Extracting Value

Lastly, by understanding the value of the information, agencies can make better decisions to drive their mission forward. Sixty percent of records, according to the Association for Information and Image Management, are unstructured, meaning that they are providing little value. And, it is estimated that organizations use only 5 to 10% of their overall data.  An example of a way for agencies to extract value is to access information that may be available on old media that can be recovered and restored, then harness the power of the information to gain the insights they need to improve operations. Another example includes using AI to extract value out of information used to feed a current workflow, such as being able to pull unstructured data from forms or records that would otherwise require a manual process.

Artificial intelligence capabilities can be the driver behind this third area.

AI and RIM

Incorporating AI into the information lifecycle management function enables agencies to classify and extract information once, then reuse downstream; seamlessly integrate content types—from physical to digital; derive actionable insight from “dark data” (information collected and stored, but never used); as well as ensure the information is managed according to policy.

Agencies should consider using AI with machine-learning capabilities to automatically classify, extract and enrich physical and digital content. ML-based classification of an agency’s physical (paper, tape) and digital (application-generated, human-generated) information adds structure, context and metadata to information to make it more predictable and usable. The resulting enriched content can then enable enhanced automation in terms of governance and workflow across the agency.

This can be accomplished by:

  • Applying neural networks and deep learning to deliver high-impact outcomes. 
  • Delivering imaging through ML/AI classification and visualization capabilities.
  • Enriching ML-based classification to unlock dark data.
  • Integrating analytics to derive insights out of the enriched data.
  • Applying and enacting retention, privacy and security policies to reduce risk.
  • Incorporating workflow automation. 

Ultimately, incorporating AI with a comprehensive information lifecycle management approach allows agencies to ingest multiple document formats into a single system, apply ML algorithms at the appropriate point, add or replace those algorithms as necessary, and offer deeper insights to achieve greater efficiencies and reduced risks. 

Government agencies are generating significant amounts of information and records in both physical and digital formats. In order to enable agencies to better leverage all this information to support the mission, they need to consider an overall information lifecycle management approach that includes comprehensive AI and ML capabilities.

Sue Trombley is the managing director of global engagement for Iron Mountain.