Judging by the results of a new survey, machine learning software is rapidly becoming the records manager’s best new friend.
The global study of opinions by 195 decision-makers conducted by AIIM found that 81% consider deep learning and machine learning technologies to be key to their future technology and business plans. But organizations are still feeling their way around with these new technologies; 87% classify themselves as “just looking” at the technology while 41% consider themselves to be early adopters. Only 13% say they’ve had machine learning models in production for more than five years.
The research, which was developed in partnership with Iron Mountain, covered a wide range of geographies and industries. While a sizable minority of respondents were in IT organizations, the results also covered line-of-business, security and records management personnel.
The results make it clear that machine learning is finding a home in content management applications, with 71% of respondents saying they have significant interest in using the technology for records management and content integration. In fact, three of the top four uses of the technology respondents favor related to document management. These include text analysis for better content classification and metadata assignment (cited by 35%), automatic understanding of context and automatic identification and indexing of documents.
Machine learning is a subset of artificial intelligence that uses computers to find patterns in large amounts of data in order to make predictions or unearth insights that humans would be unable to see. Deep learning is a subset of machine learning that mimics the human brain for applications like recognizing faces or understanding speech. Both are useful in different document-related contexts.
While security uses like risk assessment and fraud detection remain the top applications of machine learning, content-related uses are a very close second. When you compare the results to a similar survey conducted five years ago, the contrasts are striking.
For example, the percentage of organizations using automation to manage non-important content more than quadrupled to 28% from 6% in 2013. The share of organizations doing automated pre-migration data selection or metadata mapping nearly tripled to 22% from 8%.
Organizations are also using technology to mine value out of data that was previously all but unusable for analytical tasks. They face a huge backlog of “undigested” content, such as images, video and email messages that don’t easily fit into structured data formats. But 87% believe that machine learning can be applied to resolve this problem. And 82% said the ability to convert data from unstructured to structured form is very important to the success of their machine learning projects.
Companies are already at work unlocking value of unstructured information. A significant minority of respondents are already converting paper documents and semi-structured documents like invoices to usable digital form. About one-third are interpreting scanned images, classifying text or email messages and converting physical documents to tape.
Perhaps the biggest takeaway for information managers comes in responses to the statement “machine learning will revolutionize how we approach the task of information governance.” Eighty-four percent of respondents agreed. Considering that few people even knew what machine learning was five years ago, that’s a remarkable indication of how exciting the technology has become.
Digital transformation starts with replacing analog information and paper-bound processes with digital data and automated workflows. The study makes it clear that early adopters are well aware of the potential of machine learning to revolutionize information governance. Now is a good time to make sure this technology is on your radar.