The emerging of cognitive technologies have the potential to revolutionize the public sectors Many government agencies are already...
AI-based technologies include machine learning, computer vision, speech recognition, natural language processing, and robotics;1 they are powerful, scalable, and improving at an exponential rate. Developers are working on implementing AI solutions in everything from self-driving cars to swarms of autonomous drones, from “intelligent” robots to stunningly accurate speech translation. The AI could eventually revolutionize every facet of government operations. For instance, the Security, Legal and Immigration and Services in creating a virtual assistant platform like EMMAAI, that can respond accurately to human language.
Cognitive technologies are already having a profound impact on government work, with more dramatic effects to come. AI-based applications could potentially reduce backlogs, cut costs, overcome resource constraints, free workers from mundane tasks, improve the accuracy of projections, inject intelligence into scores of processes and systems, and handle many other tasks humans can’t easily do on our own, such as predicting fraudulent transactions, identifying criminal suspects via facial recognition, and sifting millions of documents in real time for the most relevant content. The potential is vast. AI will help business to increase speed, enhance quality, and reduce costs at the same time, but cognitive technologies offer that tantalizing possibility.
AI presents governments with new choices about how to get work done, with some work fully automated, some divided among people and machines, and some performed by people but enhanced by machines. In this study, we offer a roadmap for government leaders seeking to understand this emerging landscape. We’ll describe key cognitive technologies, demonstrate their potential for the government, outline some promising choices, and illustrate how government leaders can determine the best near-term opportunities.
In May 2017, Congress established the bipartisan Congressional Artificial Intelligence Caucus, and members have since introduced numerous pieces of AI legislation. More recently, the administration launched the American AI Initiative through a February 2019 executive order, and the Department of Defense released its own strategy on how to incorporate AI into national security. As government use of AI evolves, agency leaders will look for pathways to capitalize on opportunities, and the workforce will need new technical and social skills to succeed in AI-augmented workplaces. At the lower end of the scale, automating tasks regularly performed by computers could free up 266 million U.S. federal government working hours annually, potentially saving $9.6 billion. At the higher end, as many as 1.1 billion working hours could be freed up every year over the course of the next five to seven years, saving $37 billion, as estimated in a recent Deloitte Consulting LLP report on AI-augmented government.
The report produced by the IBM Center for the Business of Government and the Partnership for Public Service, the report addresses how government can best harness AI's potential to transform public sector operations, services, and skill sets. The report draws on insights from a series of roundtables with government leaders to explore pressing issues surrounding AI, share best practices for addressing solvable challenges, and work toward an implementation roadmap for government to maximize the benefits of AI. More specifically, it finds that AI could enable agencies to fulfill their numerous roles efficiently and effectively by reducing or eliminating repetitive tasks, revealing new insights from data, driving better decision-making, improving customer service and enhancing agencies' ability to achieve their missions. AI could help employees focus on core issues related to their agencies' missions and spend fewer hours on other administrative duties.
There is a “Three Vs” framework created to help government agencies assess their best opportunities for investing limited resources in AI technologies. The framework helps enable decision-makers to gauge the extent to which AI may be viable in the near future, whether there’s value in assigning specific tasks to machines, and whether AI applications are vital to tasks involving information mining and analysis.
Viable. Some tasks that require human or near-human levels of speech recognition or vision can now be performed automatically or semi-automatically using technology. Examples include initial telephone customer contacts and the processing of handwritten forms. Cognitive technologies, meanwhile, can make predictions based on large quantities of unstructured data, identify fraud patterns and clues buried in financial information, and spot trails behind public health crises.
Valuable. Just because tasks can be automated doesn’t mean they should be. Some manual functions are already performed efficiently and competently and are not necessarily attractive candidates for automation. However, it makes sense to automate functions that can be easily monitored—and thus turned over to machines—or those involving massive volumes of information. Such tasks might include determining program eligibility, processing invoices, or tabulating tax data. Moreover, professionals frequently perform responsibilities that may not actually require their expertise so AI could free up their time to perform higher-value tasks. Accountants, for instance, may analyze hundreds of contracts looking for patterns and anomalies—likely relying more on reading than accounting skills. AI technologies could take over the processes of scanning and extracting contract terms. In fact, cognitive technology in the legal field can find relevant documents for discovery faster and more thoroughly than lawyers do.
Vital. Processing high volumes of certain business transactions in government, such as those requiring a high degree of human attention and analysis, may not be achievable without the support of cognitive technologies. For example, with the help of optical character recognition, one Georgia agency processes 40,000 campaign finance disclosure forms per month, many of them handwritten. Machine learning could be critical to numerous government functions, from fraud detection to cybersecurity. A learning system that can respond to ever-changing threats by learning from past experience and external modeling may be the best defense against adversaries ranging from rogue states to cybercriminals.
Four Ways to Deploy AI in Government
Relieve. This approach lets technology take over mundane tasks, allowing workers and employees to focus on higher-value tasks. In the U.K., one central agency automated the most tedious aspect of its call center work—opening case numbers. The agency estimates this reduced handling times by 40 percent and processing costs by 80 percent.
Split up. Automation technologies can be applied to a specific job or task, leaving humans to complete the rest and perhaps supervise only the work of the application. For example, at the United Nations, there is language translation software that creates live transcripts during assembly meetings for spectators, while human translators could revise them later for the publication. In addition, the White House and U.S. Citizenship and Immigration Services, have designed chatbots to answer some basic online questions while leaving more complicated queries to humans.
Replace. In this model, AI can be used completely to replace entire functions or the job once performed by humans. The best opportunities include repetitive tasks. For instance, the U.S. Postal Service uses handwriting recognition technology to sort out mail by ZIP code; (the work that belonged to a mailman ) some machines can process up to 18,000 pieces of mail an hour, in surpassing human with far extent.
Augment. Human workforce and skills can be combined with AI technologies to achieve faster and better results. When technology is designed to augment, humans who still remain in the driver’s seat. An example is IBM’s Watson for Oncology, which uses cognitive technology to recommend individual patient treatment plans to physicians, citing evidence and a confidence score for each recommendation to help doctors make more fully informed decisions.
Government employees in the future will need new skills to succeed in an AI-enabled world. As AI becomes more ubiquitous in everyday business government workplaces should emphasize expertise in technical, digital and data literacy.
- The report recommends three paths for agencies: Sufficient funding for AI projects and basic machine learning skills to the government employees in extensive and ongoing training about technology, digital skills, and data analysis in order to succeed in an AI workplace.
- The stakeholders should work with other relevant agencies and academic institutions to establish a team for AI talent similar to the U.S. Digital Service, governed by rules that make it easy to hire top AI talent from the private sector for time-limited stints in government to help the state department or agencies that need AI expertise.
I will likely fundamentally transform how government works, and the changes may come sooner than many expect. As cognitive and automation technologies advance in power and capability, government agencies can bring more creativity to strategic workforce planning and work design, and leaders can work together to analyze the interplay of talent, technology, and design to propose a path forward for AI in government.
Other contributors to this article: Claude Yusti, William D. Eggers, David Schatsky, Dr. Peter Viechnicki, Tatiana Sokolova, and Alayna Kennedy from IBM, and Peter Kamoscai, E.A Nambili Samuel, and Katie Malague from the Partnership for Public Service.