AI in government

The use of artificial intelligence (AI) in government refers to the implementation of AI in governmental affairs and the rules and regulations those officials make to legislate how private companies and individuals use it.

Like other advanced technologies, AI and automation not only have the potential to improve the lives of citizens around the world but also introduce significant risks.

Government use of AI is a controversial topic, given the power it has and how it might be misused to benefit some and penalize others. However, governments are heavily invested in exploring AI technologies out of both opportunity and risk. There is an opportunity to use AI to improve their citizens’ lives and grow the economy. The inherent risk with AI is that other countries might use it to become more adept at war and economic growth. In certain zero-sum scenarios, governments that excel at AI might put other countries at a disadvantage.

From the federal government to local governments, every elected official is curious about how AI can help them do their jobs better. Many governmental leaders believe that embracing and mastering AI will provide a competitive advantage against fellow nations and protect them from potential conflicts.

Government agencies have access to several rich datasets of structured and unstructured data that the adoption of AI can help to provide more insights. Understanding that data can help governments provide more efficient services to their citizens.

Every government can embrace trustworthy AI, which provides guidance on the safe and secure deployment and use of AI systems.

There are over 1,000 AI policy initiatives from 69 countries, territories and the EU, including almost 800 governance initiatives, according to the Organisation for Economic Co-operation and Development (OECD) (link resides outside of IBM.com).¹

While each country is pursuing its own AI strategy, there are inherent risks in AI development that might impact every country. Therefore, there are several global partnerships where leaders assemble to discuss those challenges and share resources.

For example, the UN convened an AI advisory council (link resides outside of IBM.com)² in 2023 that included governmental officials, private sector leaders and academic researchers. It also holds an annual AI for Good Summit. The World Economic Forum (link resides outside of IBM.com)³ also has an AI task force.

Here are countries or regions that are approaching AI:

• United States
• China
• European Union
• India
• Japan
• Brazil

The United States government has invested funds and research into the emerging field of AI.

In 2023, US President Joe Biden issued an executive order (EO) (link resides outside of IBM.com)⁴ that requested federal agencies participate in developing guidelines, standards and best practices for AI safety and security. The White House Office of Management and Budget (OMB) announced at the end of November 2024 that progress continued to be made for first government-wide policy (link resides outside of IBM.com)⁵ to harness AI and mitigate its risks based on this EO.

The EO is part of a broader goal of the US government to establish smart AI governance to take advantage of any opportunities and prepare for challenges while staying ahead of other countries.

The American government is only one of many global governments that is investing in AI. Industry experts often describe the battle for AI supremacy to be a race between China and the US (link resides outside of IBM.com).⁶ China is investing heavily in developing AI tools. The Chinese government has also developed laws on how its citizens can use generative AI tools such as ChatGPT (link resides outside of IBM.com).⁷

The United Kingdom established an Office for Artificial Intelligence (link resides outside of IBM.com)⁸, which was later folded under the Department for Science, Innovation and Technology in February 2024.

It also announced in 2021 a 10-year plan (link resides outside of IBM.com)⁹ to make Britain a global AI superpower through investments in research and development, wider adoption of AI in various industries and regions and increased investment in AI companies.

The European Union announced Europe for the Digital Decade (link resides outside of IBM.com)¹⁰, “a human-centric, sustainable vision for digital society throughout the digital decade to empower citizens and businesses.” It intends to increase research and industrial capacity while in an ethical way that protects citizens’ rights.

EU member states combined to create a Coordinated Plan on Artificial Intelligence  and regulatory framework proposal to add more rigors to their AI approach.

India’s AI policies have changed over time. At the 2023 G20 summit (link resides outside of IBM.com)¹¹, the government asserted it “will pursue a pro-innovation regulatory/governance approach that maximizes the benefits and considers the risks associated with the use of AI.”

It has also actively encouraged the use of AI for social welfare (link resides outside of IBM.com)¹², specifically in diseases detection and agricultural improvements.

According to the Economist (link resides outside of IBM.com)¹³, the ruling party, the Liberal Democratic Party (LDP), wants Japan to become “the world’s most AI-friendly country.” The country allows corporations to train their models on all types of content without restrictions and does not have any specific laws governing responsible use of AI. Despite these lack of restrictions, it is not considered among the leading countries in AI development (link resides outside of IBM.com)¹⁴. But it hopes it will catch up in the next few years.

The Brazilian government is taking steps (link resides outside of IBM.com)¹⁵ to boost AI leadership in South America. It recently announced a multi-billion-dollar campaign to invest in homegrown AI technologies to reduce its dependency on external tools.

• Establish clear governance
• Appropriately fund AI research and usage
• Explore private-public sector and academic partnerships
• Ensure compliance

Federal, state and local governments can establish clear governance policies on how they plan to use AI. They can specify in which contexts they should use AI and when its use is not acceptable. The government also provides guidance to individual states, though they might make their own laws regarding AI use.

Historically, governments have funded technology research to kickstart innovation. The Internet is indebted (link resides outside of IBM.com)¹⁶ to the American Defense Advanced Research Projects Agency, which researched network connectivity. Experiments and research at NASA created several modern inventions we still use today, including LEDs and memory foam (link resides outside of IBM.com)¹⁷. The US government can research AI solutions and fund other researchers who are looking into practical uses for AI.

In many cases, the private sector is leading the development and deployment of AI tools and systems. While the private sector can develop their own tools, it’s clear that most governments will need to adopt some technologies built by companies. The US has started exploring a “Manhattan Project-style initiative” (link resides outside of IBM.com)¹⁸, which would establish a closer private-public collaboration and fund AI development. In addition, leading universities and nonprofits are conducting significant research into AI, and governments might benefit from partnering with them.

Once policies are in place, the government body responsible for enforcement must ensure everyone is following the rules. Compliance relates to the government agencies themselves and private companies within their jurisdiction. Examples of compliance include establishing and following safety guardrails and other laws dictating the use of AI to protect privacy and avoid any type of unequal application. For example, a bank or real estate company cannot use algorithms that discriminate against a specific group for housing or loans. Governments might need to audit these tools to ensure compliance.

Several types of AI technologies are deployed in government to power many different AI use cases.

• Computer vision
• Generative AI
• Intelligent automation
• Machine learning
• Natural language processing
• Optical character recognition

Computer vision uses machine learning and neural networks to help computers parse information from images, videos and other visual inputs and turn it into actionable steps. A core component of computer vision is pattern recognition where computers can identify similarities between objects to make decisions on what is on screen. Governments can use computer vision for several key functions. One example is facial recognition, which can help identify suspects in criminal cases. It can also be used to make logistics decisions, such as analyzing traffic patterns to determine if a stop sign or traffic light is needed in a specific area.

Generative AI impacts all fields, especially governments and their relationship to their citizens. Governments use generative AI in several instances. For instance, government agencies have begun to introduce the technology in their internal tools to help public employees perform their jobs more accurately and efficiently. IBM helped the City of Helenski deploy virtual assistants that could help busy employees answer constituents' questions quicker and more accurately.

Gen AI tools can help those employees locate relevant departmental information quicker and also ask questions of virtual assistants to understand their options. Secondly, those same agencies can introduce public-facing chatbots that citizens can query to get information.

These are especially helpful for organizations that directly interact with citizens. AI-powered chatbots are also a powerful tool for congresspeople to interact with their constituents. The third use of generative AI is to help governments modernize their applications and their code. Gen AI can help connect modern code to legacy technologies, such as mainframes, to ensure they run well.

An IBM Institute for Business Value study found that the majority of citizen respondents across the globe agreed that governments should use gen AI for “customer service, tax and legal advisory services and for educational purposes.”

Intelligent automation is the use of automation technologies such as business process management (BPM) and robotic process automation (RPA) to improve decision-making and actions across the government. Governments can use intelligent automation to process applications for citizenship or grants, for instance. It can also help hire employees quicker by identifying candidates that have the right skills listed in their resumes.

In addition to powering computer vision, machine learning accelerates and improves upon other AI technologies producing powerful outputs. It helps collect and analyze data more efficiently, offering government agencies the opportunity to create more actionable insights from the data they collect.

Natural language processing (NLP) is a technology that teaches computers to understand and communicate with human language. It is a core component of generative AI, providing the ability for chatbots and virtual assistants to understand prompt commands and take action. NLP can help government officials sort through submitted information, such as requests for assistance, and find the most pressing issues to address. It can also help government agencies sift through social media and other online forums (link resides outside of IBM.com)¹⁹ to understand what issues their constituents are having.

Optical character recognition (OCR) is a technology that converts images of text into a machine-readable format. This is the technology that helps digitize content in the Library of Congress to establish searchable databases and create redundant backups in case of document loss or destruction. It is also used by other government agencies to bring online their historic documents.

There are several areas of government services that benefit from AI. These include:

• Coding and legacy modernization
• Cybersecurity
• Disaster preparedness
• Healthcare
• Law enforcement
• Procurement

Governments can use AI to identify bugs in existing code and modernize it to newer languages. For example, the Office of Personnel Management (OPM) plans to use AI to modernize its legacy IT system to better deliver services to its retirees.

While AI itself can create risks for government infrastructure, governmental agencies can also use AI to improve their cybersecurity defenses. Studies show that using AI and automation capabilities can detect and respond to cyber incidents with greater speed. AI threat detection and mitigation can help governments respond quicker to safeguard important datasets.

Climate risk and other geopolitical challenges require national governments to improve their preparedness for upcoming incidents. AI can help governments look at historical weather and current environmental data to better predict potential issues such as floods, hurricanes or wildfires. IBM has worked directly with NASA to convert NASA’s satellite observations into customized maps that can track environmental changes such as natural disasters.

AI can help unlock better outcomes for patients while reducing costs (link resides outside of IBM.com)²⁰. Healthcare agencies specifically can use AI to analyze data from drug trials to expedite drug discovery and approvals. They can also use it to minimize manual tasks associated with paperwork and approval processes. It can also collect and analyze public health information to better detect potential pandemics or other health issues.

Powerful ML platforms can help federal agencies concerned with law enforcement to better track threats in and outside of the country and solve crimes. Using pattern recognition, AI can analyze surveillance cameras, such as at airports, to identify suspects. However, governments must use this technology fairly and avoid discriminatory behaviors to protect their citizens’ civil liberties. It can also help legal departments expedite cases. The German government worked with IBM to use AI to improve its ability to work through its backlog of cases. IBM customized an AI bot named OLGA that extracted metadata and provided case categorization.

Many federal governments are the largest purchasers of goods and services within their borders, and many have specific rules in place that dictate where and from whom they can purchase those items. AI procurement can collect information about potential suppliers and help governments select the right organization based on their stated criteria.

There are several benefits for governments that use AI.

• Create better predictions
• Improve citizens' quality of life
• Stimulate economic growth
• Streamline decision making

Governments that use AI can have more powerful predictive analytics that help them with important tasks such as external threat detection, health crises and financial issues like inflation. By understanding what is likely to happen quickly, governments can make smarter decisions that might minimize the effect of these issues.

One of the main reasons why governments have embraced AI is that they believe it can help them improve efficiency and provide greater services to citizens. Cutting down manual work to collect and parse data, approve requests and answer citizen questions all have a knock-on effect of improving their lives. For example, AI can make collecting information for legal documents and processing and shipping out items, such as passports, quicker.

Using AI helps economies in several ways. First, the increases in efficiency drive down government costs, which can either be passed onto consumers or used to pay down the national debt. Second, government investment in AI means that private AI and technology companies might receive grants or otherwise sell their services to the government. Third, companies that want government business will also need to embrace AI, further creating more economic opportunities for those companies providing AI services.

AI can improve government operations by providing policymakers with more information and the ability to query generative AI to understand potential strategies. Governments can use AI models to decide budgets quicker and allocate the funds to the agencies and nonprofits that depend on them.

The responsible use of AI is integral to any government overcoming the challenges the technology presents.

• AGI
• Biases and prejudice
• Data breaches
• Inequality of services

Most, if not all, AI experts believe humanity will be able to create artificial general intelligence (AGI). This theory is that AI will eventually reach the level of human cognition, effectively being able to do anything humans can do and, often, without human operators. There are many incredible benefits from AGI,  such as every human on earth having access to his or her own 24-7-365 AI doctor who can offer diagnoses in real-time and help prevent future illnesses.

There is also the fear that the AI becomes too powerful or quickly evolves into superintelligence where machines start making decisions that benefit themselves and not humanity. While AGI and superintelligence are not expected to happen anytime soon, and might never happen at all, it demonstrates why governments must be closely involved in AI development and regulation.

Citizens expect their democratically elected governments to protect their rights and eliminate discrimination. Because AI models stem from human-made data, they can inherit those human biases and prejudices. Government agencies must establish their own safety protocols to protect their citizens’ human rights and minimize any harm from AI to its citizens.

For example, AI-powered predictive policing tools rely on historical data. If that original data was the result of unfair racial profiling, it can perpetuate a lasting negative effect of disproportionately targeting minority communities.

Hackers accessing data threatens the uptime of public services and, for the federal government, national security. The use of AI applications can create greater exposure of risks. At the same time, governments can use AI to protect their overall data.

For some citizens, AI is a simple tool they use in their daily lives to get answers and perform tasks more efficiently. For example, a local resident who wants to understand how a new law will affect them might prefer to interact with a chatbot. However, there are others who would prefer calling and talking to a government agent. As governments embrace AI and build tools that use it, they also must meet the needs of citizens who do not want to use them.