The public sector is the backbone of our daily lives — maintaining roads, educating our children, ensuring national security, and providing essential social services.

Its various government agencies – at national, regional, and local level – have to process a vast amount of disparate information about millions of citizens, as well as formulate complex policies and ensure they are carried out and enforced. And they often have to do all this with limited budgets, while adapting to political and economic change.

It’s a profile that suggests generative AI (GenAI) could have a big effect in the coming years on the way governments worldwide work, from the national level all the way down to local councils. Let’s explore the benefits and the top use cases for secure AI in government and the public sector.

Benefits of secure AI in government

Mckinsey estimates that GenAI can have a $480 billion productivity impact on the public sector and adjacent industries in the coming years. Boston Consulting Group sees those gains rising to $1.75 trillion annually by 2033, with the government market for GenAI products projected to grow 50% per year.

Yet, the path forward isn’t without hurdles. Government agencies must navigate concerns around security, accuracy, and public trust. These organizations handle sensitive data, where mistakes or leaks could have dire consequences. It’s no wonder most are treading cautiously. But for those ready to harness GenAI responsibly, the opportunities are enormous.

The UK government recently published a comprehensive plan to harness AI, saying the technology could be the “single biggest lever” for the government to achieve its key goals, including stronger economic growth, improved healthcare provision, and becoming a leader in clean energy.

“AI should become core to how we think about delivering services, transforming citizens’ experiences, and improving productivity,” the report said.

Canada published its own guidelines last year, encouraging federal agencies to start using AI for low-risk use cases such as writing emails and drafting documents.

Let’s look at three transformative ways GenAI can reshape the public sector, offering insights into how governments can seize this moment.

Better, faster citizen services

Anyone who’s had to sit in waiting rooms for hours or plow through reams of paperwork to apply for licenses, permits, or social benefits, or just to access relevant information, knows this is an area for public-sector improvement.

Some 60% of callers to the UK tax authority, for example, had to wait more than 10 minutes in the year through April 2023 , a proportion that has risen every year since 2018. In the United States, callers to the Social Security Agency’s national number faced an average wait time of 16 minutes in 2020, up from just 3 minutes in 2011.

This is where AI-powered agents, drawing on government knowledge bases, can have a big impact in cutting wait times, improving satisfaction with public services, and freeing up workers for higher-level tasks. Public agencies can look to the private sector, where the customer-service use case for GenAI has already shown proven success.

After securely verifying a user’s identity, an AI assistant could – in multiple languages – provide personalized information and answer questions about their benefits, such as their eligibility, required documentation, or payout calculations.

The technology is already being used to help buses find faster routes, and to speed up responses to emergency calls. In America, 911 is experiencing an emergency of its own. For every four people needed to answer 911 calls, there is one office chair that sits empty. Police departments and other emergency services throughout the country are taking longer to respond to the 240 million calls per year, half of which aren’t emergencies.

Several U.S. cities, including Phoenix and Fort Myers, FL, have started using AI to triage emergency calls. Their solution integrates with 911 dispatch and electronic health records to filter low-urgency calls out from more urgent cases. In the future, an AI assistant integrated with emergency apps or text services could gather key details from callers and relay them to dispatchers instantly, reducing the time spent on initial information collection.

Strengthening national security

Military and intelligence services have been relatively early GenAI adopters, realizing that the technology could give them a crucial edge in gathering information and obtaining actionable insights from it.

Imagine the following scenario. A country’s military is scrambling to come up with a plan to airdrop crucial supplies to ease a humanitarian crisis in a remote part of the world. Traditionally, that response might involve gathering multiple intelligence sources, such as human intelligence (HUMINT), satellite imagery, and news reports. Inter-agency meetings and approvals at multiple levels are needed to formulate and launch the operation.

GenAI’s ability to synthesize large volumes of external and internal data in real-time could streamline this process significantly. It could integrate live data such as weather information with historical context to generate options on the best operational approach and timing.

The U.S. Army War College is developing AI tools that provide direct access to accurate and timely data to support operations and augment human planning. During one scenario depicting a humanitarian crisis, the AI-enabled “strategic advisor” calculated vehicle travel times and provided recommendations to enhance logistical efficiency.

And there are a myriad of less dramatic, but no less useful, ways in which GenAI could improve military productivity. Just like large private enterprises that are seeing benefits from GenAI, militaries have massive stores of data held in diverse formats and face intense logistical complexities in coordinating troops, supplies, and equipment around the world. Much of their documentation is confidential and heavy in highly specific jargon, making it essential to have a privately deployed large language model (LLM) that has been trained on their data.

Simplifying policymaking

The process of policymaking, whether planning local road works or creating reports on national economic development, is complex, manual-intensive work encompassing data collection, stakeholder consultations, impact assessments, and multiple revisions. Even expert policymakers stand to benefit from GenAI’s ability to scan millions of documents and summarize key findings and trends, in addition to modeling potential policy outcomes.

Agentic AI tools will take this a step further, giving policymakers the ability to automate further functions such as plotting out graphs to visualize text-based data. In housing policy, for example, an agentic tool could identify regions with housing shortages, propose zoning rule changes, simulate the impacts, and then generate tailored policy recommendations.

Rather than taking over decision-making from expert policymakers and elected officials, the power of GenAI will be in assisting the research process, assessing options, and running simulations so that decision makers aren’t starting from a “blank page” as they often are now.

Adoption through building trust

Overall, most public agencies are unlikely to race into GenAI adoption before they’ve been sufficiently reassured by its results and safety record in other sectors. Any AI failures or inaccuracies, such as the New York City chatbot that reportedly gave erroneous advice to citizens, can have a damaging effect on public trust.

Critical to adoption is building secure AI systems and working with partners that offer transparency and auditability. Cohere recently published the Secure AI Frontier Model Framework which enables critical infrastructure organizations in the public sector to build safe and secure solutions for their customers with the goal of advancing the safety and security of our AI products.

As GenAI goes more mainstream and public agencies work with trusted AI partners to ensure the right safeguards are in place, the huge potential impact on government productivity and costs will be increasingly hard to ignore.

Mastering private AI deployments

Is your business ready for a secure, private AI deployment?

Download the eBook

Secure AI in government: Improving services for all