The real value of artificial intelligence for intelligence analysis: Humans can do more

The benefits of artificial intelligence (AI) go far beyond saving time. After all, intelligence work is never over, there is always another issue that needs attention. Therefore, saving time with artificial intelligence will not reduce labor or intelligence budgets. Instead, artificial intelligence's greater value comes from something that could reap an "automation dividend": intelligence analysts can put more of their time to better use after the workload has been eased by these technologies.

In fact, studies of industries ranging from banking to logistics show that the greatest benefits of automation come when human workers use technology to "move up the value chain." In other words, they spend more time performing tasks that are of greater benefit to the organization and customers. For example, when automation frees supply chain workers from tasks such as measuring inventory or filling orders, they can reallocate time to create new value by matching specific customer needs with supplier capabilities. In terms of intelligence analysis, using artificial intelligence to immediately extract hard-to-find signs and early warning clues from chaotic data can allow intelligence analysts to perform more valuable work to determine whether a given early warning clue represents a real threat.

There are two main ways that additional time can be used to create additional value: intelligence analysts can spend more time on high-value tasks they are already completing, or they can add new high-value tasks.

Do More: Human Focus on Human Tasks

However, before these benefits can be realized, intelligence organizations must determine which are the most valuable tasks and which are best suited for human workers to perform. First, let's compare humans to computers or other machines.

The key is to understand the difference between specialized intelligence and general intelligence. Even a simple pocket calculator can outperform the best mathematical abilities in some tasks. But, while fast and precise, it's the only task a pocket calculator is capable of performing. It has a very narrow, specialized intelligence. On the other hand, humans tend to outperform even the most advanced computers in general intelligence. Thomas Malone, a professor at the Massachusetts Institute of Technology, explained, "Even a 5-year-old child has a higher intelligence than the most advanced computer programs available today. A child can have more intelligent conversations, discuss a wider range of topics, and operate more effectively in unpredictable physical environments."

So while machines are better than humans at crunching large amounts of data or working to extreme precision, humans are better at tasks that vary dramatically with context or involve high levels of human interaction. Human workers and artificial intelligence tools can work together to play to their respective strengths; artificial intelligence crunching massive amounts of data, humans handling highly variable tasks. Within intelligence organizations, human analysts can move up the value chain by offloading many repetitive data processing and development-related tasks to machines. They can then devote more energy to analyzing, planning, and directing tasks that often require more creativity, communication, and collaboration with colleagues and decision makers.

Our model makes similar predictions for intelligence analysts. Using artificial intelligence for tasks such as data cleansing, labeling, or pattern recognition allows all-source analysts to spend more time performing context-sensitive or uniquely human tasks. As a result, analysts of the future are likely to spend more time collaborating with others—58% more than today.

How can greater collaboration be brought into play throughout the intelligence cycle? For example, during the dissemination phase, analysts provide information to decision makers, working with them to enable them to make the best decisions. If artificial intelligence can take on more of the preparatory work of gathering resources, creating graphics, and even drafting reports, human analysts can focus on the needs of decision makers and the implications of the situation. In this scenario, the analyst simply provides Al with the topic of an upcoming briefing or finished product. From there, Al can automatically generate a list of relevant reports to read, pre-select maps or images, mark brief descriptions of relevant features, and even write brief background event summaries.

Intelligence agencies must determine which are the most valuable tasks and which are best suited for human workers.

A similar shift has already occurred in journalism. Artificial intelligence is being used to automatically generate a simple Washington Post, with bots publishing 850 articles in its first year, covering everything from the Olympics to the election. The Associated Press found that by automating detail-oriented tasks like writing company earnings reports, the use of robots can reduce journalists' workload by 20%, allowing them to focus on reducing errors and spotting larger trends. As a result, errors in corporate earnings reporting have decreased, even as production volumes have increased. Intelligence analysts could also benefit from a similar arrangement: Artificial intelligence could generate routine intelligence summaries or daily reports, allowing analysts to focus on synthesizing these reports into larger trends, or tailoring reports to the preferences of specific decision makers.