You are here

March 26, 2018

In a previous post on Thinking Next, I kicked off a series on artificial intelligence (AI) and machine learning by looking at the surprisingly long history of those technologies. In that article, I explained how AI has evolved into one of the most buzz-worthy innovations of today, and discussed the technologies that were opening the door for the AI revolution.

From there, we heard from two very exciting and innovative emerging technology companies that are each using AI to accomplish amazing things within the enterprise. The first was Astound, which is using AI to help make constituent and employee service more efficient and effective. The second was Cylance, which spoke to us about how AI can help secure networks and fight cyber threats.

These are two exciting companies doing exciting things. But they’re two of hundreds of AI solutions either on the market or making their way into the market today.

And that’s something that the federal government should be extremely excited about. Why? Because AI has the truly transformative power to revolutionize how government works and to help our government and its agencies overcome some of the largest challenges and hurdles keeping them from accomplishing their missions.

Let’s take a look at some of the overarching challenges impacting all of the federal government and how AI can help to solve them.

So much to do, so little time
If there is one universal challenge impacting every organization, agency and entity within the federal government, it’s having to accomplish a large, essential mission with less than optimal resources. This is exasperated by constant demands to do more, be better, and deliver more services to constituents.

And—in the midst of being asked to do more—hiring freezes, budget cuts, and staff reductions mean that the government has to do more with less. It’s a problem that’s compounded by the wave of federal employees reaching or approaching retirement age and the federal government’s ongoing struggling to recruit and retain top talent away from the private sector.

Then there’s data. The military and federal government has a lot of it. In fact, you could argue that they have too much of it. This data can be unstructured data from social networks, it can be ISR data and video from military sensors, or GEOINT data from satellites. Regardless of what kind of data it is, and where it came from, it all holds the promise of making the government and military more proactive, more prescient, and more effective.

But for that data to be helpful, it has to be processed, exploited, and distributed to those who can harvest its intelligence effectively. But analyzing the mountains of data that is available to the government—even just to see if it’s of value—can be more work than the government has warm bodies to handle.

So what do these challenges have to do with AI?

Each person, empowered
When resources are tight and agencies have to do more with less, it’s essential that employees are operating effectively and only handling the most critical of tasks. AI can enable this.

AI is a learning technology. All tasks or processes that have past information or data can be used to teach AI systems. Once they have been taught, AI systems can then automate that process or task. This is something that can be effectively rolled out across agencies that provide benefits or services to citizens.

By providing information from all previous constituent interactions, an AI system can learn how to adjudicate benefits more quickly and effectively than a human can. There would still be a human in the loop for some hard cases that don’t have precedent or require a judgement call, but the easier ones can be automated.

In this scenario, both the agency and the constituent can benefit from AI. The agency gets to keep their limited human resources focused on more difficult cases and tasks that are outliers. This means that they’re effectively free from redundant, simple work to focus on harder, more mission-critical tasks. They’ve had some of their workload automated so that they can focus on what can’t be automated. The benefit to the constituent is more rapid access to services and less red tape.

But adjudication of citizen services and benefits is just one area where AI can automate work and even perform at a higher level than human employees. Another is identifying fraud, waste and abuse.

Whenever there are citizen services and benefits available, there are always going to be individuals that attempt to game the system and fraudulently access government funds for their own advantage or profit.

This can be medical professionals that bill for extra services that weren’t rendered just to claim additional reimbursements. It can be individuals trying to access SNAP funds for products and services that are outside of their original, intended use. Or, this can be tax fraud—where individuals try to cheat the IRS by falsely reporting their earnings. Regardless of what form it takes, fraud, waste and abuse takes taxpayer money away from government coffers and keeps them from being spent on those that are actually in need.

Luckily for the government, AI can assist in this area as well.

Activities that involve fraud, waste and abuse often illustrate repetitive patterns that can be difficult to spot for humans, but relatively easy to identify for AI systems that have been trained to look for them. By comparing the suspicious activity of a particular beneficiary against previous fraudulent behaviors and the patterns or behaviors of the average beneficiary, abuses of the system can become very obvious to AI systems in relatively short order.

How effective are these systems? An AI-based fraud-prevention system that I helped to develop years ago at the IRS has helped to identify and flag more than $3 billion in income tax fraud each year.

Then there’s the matter of data. AI solutions are also capable of parsing through the mountains of data that the military and federal agencies are generating—helping them to identify patterns, flag data that is of value, and categorize data for easy exploitation and dissemination. This can help take mountains of data and identify just the information that’s necessary, while it’s still relevant, and deliver it when and where it’s needed most.

The federal government is short staffed. It’s underfunded. And it’s asked to do more and perform at a higher level than ever before. To continue accomplishing their missions, federal agencies need to empower their employees to do more with less. AI and machine learning is the key in these instances thanks to its ability to empower each individual, automate processes and enable employees to put their focus on the most difficult and mission-critical of tasks.

For additional information on AI and CSRA’s artificial intelligence and machine learning capabilities, click HERE.

Please note: The content on this page was originally posted on prior to its acquisition by General Dynamics. This content was migrated to on July 9, 2018.​