The U.S. government won’t miss out on the wave of artificial intelligence and machine learning (AI/ML) coming in the 2020s.
That’s not to say it will be an easy ride. Federal agencies have to figure out how to accomplish ambitious AI/ML goals within the bounds of budget, personnel and political necessity. If they get it right, they’ll hand some tasks over to AI algorithms while freeing more people to work on solving human problems beyond the ability of machines.
These are three of the most likely ways that AI will help federal agencies in the 2020s and beyond:
Predicting Likely Outcomes
If AI/ML algorithms have enough data across a long enough time span, they can scan for patterns repeatedly and teach themselves to predict likely outcomes. Because the U.S. government has some of the world’s largest repositories of data, it’s well positioned to develop predictive AI. Likely use cases:
- Vehicle maintenance. From compact cars to aircraft carriers, the federal government maintains massive equipment fleets. Each vehicle has moving parts that wear out. Predictive maintenance can tell fleet managers when to remove worn-out parts in advance, preventing breakdowns that delay critical missions.
- Safety. Federal inspectors need to detect hazards in aircraft, mines, food-processing plants and many more applications. Computer-vision technology can analyze pixel patterns in still images and video frames to detect patterns that the human eye misses. With enough data, this kind of AI can make inspections much more efficient — targeting the most hazardous sites and predicting likely hazards, which can prevent accidents.
Tightening Security and Compliance
It’s easy to imagine government agencies using AI for the espionage and secret missions that fill so many TV and movie scripts. Some of that will be happening in the 2020s, of course, but with much less pulse-pounding drama. Examples:
- Border and airport security. As computer-vision AI matures, agencies will get better at identifying and detaining potential bad actors. It may even be able to detect people trying to disguise their appearance. Along the border, computer vision can scan video feeds to detect likely entry points and give guidance on where to deploy personnel.
- Regulatory compliance. Companies submit millions of PDFs and other documents in regulatory disclosure filings. Machine learning algorithms and Natural Language Processing will get better at scanning these documents to automate processing, streamline approvals, and improve accuracy.
Automating Manual Processes
Robotic process automation (RPA) is making waves across the private sector in factories, distribution centers and other commercial applications. The federal government is also getting into RPA. Examples:
- Data entry. For decades, the government has collected data in multiple formats like PDFs, images, video, text and spreadsheets. Workflows that used to require manual processes to enter these documents into databases can now be automated, enabling the government to get more work done with limited staffing and budgets.
- Conversational AI. Virtual assistants and chatbots will help government agencies make data entry faster and more accurate. That’s part of a broad move toward voice-driven automation that boosts the efficiency and effectiveness of public services.
What’s on the Horizon in 2020
Federal agencies face constraints that private businesses rarely endure. For instance, rules requiring data privacy and public accountability give the federal government a small threshold for failure. It’s no small challenge to develop AI systems in this environment.
Nevertheless, agencies are launching AI pilot projects and inviting companies to compete for a chance to participate. As these small projects expand, federal agencies will gradually add more AI/ML to their technology portfolios in 2020 and throughout the decade.
Ultimately, the federal government and the private sector reach their destinations via wildly diverging routes, but they share the same desire — using thinking machines to unleash the superior power of human cognition.
–Varun Dogra, chief technology officer