Generative AI and the Public Sector

Angie Heise, Corporate Vice President, Microsoft Worldwide Public Sector

The potential to revolutionize how we address challenges

Artificial Intelligence (AI) is changing the game for businesses and organizations of all types. New advanced natural language processing capabilities around generative AI and large language models have captured the world’s imagination. These capabilities have the potential to revolutionize both the way governments interact with citizens and how the public workforce addresses society’s challenges. It is imperative that public sector organizations begin to think about how to leverage and responsibly employ generative AI for government and education purposes. 

Responsible approaches to Artificial Intelligence

Advances in generative AI demand similar progress in our consideration of the implications for society, ensuring that the applications of this powerful technology are responsible, ethical, broad-based in their benefits, and deployed in ways that provide value for society. The building blocks have already been laid for the principles and frameworks that govern responsible AI and will continue to evolve along with the technology.

Responsible AI standards can ensure that deployments value fairness, privacy and security, reliability and safety, inclusiveness, and accountability. But the public sector cannot remain frozen as AI changes the world around us. Communities deserve the more effective and efficient services that AI can bring to society, and public sector workers deserve the most effective and efficient tools to carry out their mission. Forward-thinking leaders will seek to engage these technologies in ways beneficial to their populace, and through engagement, learn to understand and harness their power for societal good. 

AI Public Sector Vision and Disruption

There are moments in the evolution of technology that impact society and the economy deeply – disruptive advancements in innovation whose effects are not fully felt until years after they debut. We have seen industries changed irreversibly by the internet, smartphones, and cloud, just to name a few. Although it is hardly a new technology topic, recent advancements have catapulted AI into the firmament of game changing, disruptive technologies, and it is worth taking a step back to contemplate what the long-term implications for the public sector may be in years to come.

How Artificial Intelligence could support public sector organizations

Although advanced AI has a host of implications for the public sector, here are some relevant areas of impact for consideration: 

• Citizen Services: Generative AI can help governments and public sector organizations provide enhanced service experiences that make government more accessible and less time-consuming by acting as an “Information Assistant” – answering frequently asked questions, recommending services based on inputs, and even handling simple transactions.
  
  Many governments have already experimented with chatbots to help answer simple questions about COVID vaccinations, provide support during tax time, and offer answers to common inquiries. Generative AI helps chatbots handle more open domain questions over more sophisticated and complex materials, including rapid responses to a broader range of questions at anytime from anywhere, increasing accessibility for citizens while simultaneously increasing government efficiency and reducing administrative burdens.
  
  Citizens can even provide a narrative of their current circumstances and discover service options they previously did not know existed. These tools also free up public sector workers to focus on strategic projects instead of being tied down to mundane, repetitive functions such as responding to common questions.
• Internal Efficiency: Government can be complex even for government employees! Providing public sector workers with the capacity to intuitively search and interact via chat with intranets and public sector materials in an automated fashion eases onboarding of new employees, increases efficiency between silos and departments, and minimizes administrative burdens. This capability lets public sector staff focus on their mission priorities, reducing burnout and allowing them to do more with less.
• Deep Data: Large language models can tackle the intersection between vast troves of data which may have been previously analyzed separately and manually. Simple prompts to the AI can yield both typical and unexpected connections between topics and domains that can help to spur the analytic process.
  
  Insightful and succinct summaries of vast amounts of media coverage or public feedback can be generated in seconds. Generative AI helps to objectively challenge conventional wisdom – raising new angles, questions, or counterarguments that may have been implicitly screened by the bias of the author. This approach ultimately yields a stronger and more comprehensive output.
• Creative Aid: No more writer’s block! Generative AI can provide a helpful first draft of abstracts, outlines, speeches, simple correspondence, memos, frequently asked questions, whitepapers, and citizen guides. While official communications should always require a human in the loop to verify accuracy, apply human “voice,” and ensure that the information is complete and not misleading, generative AI as a creative writing aid can accelerate the process dramatically and help light the creative spark while reducing time-to-completion for common writing tasks.

By improving citizen services, increasing efficiency, better managing and analyzing data, and serving as a creative aid, generative AI can help to create a more effective, inclusive, and responsive government.

Generative AI can also help create a more efficient, productive, and rewarding work environment for public sector employees. Governments should carefully consider the implications of using AI in their operations and take appropriate measures to ensure that the technology is used ethically and responsibly. Now is the time for public sector organizations to begin leveraging and adopting generative AI capabilities, and they can and should do so from a position of engagement and experimentation. 

Responsible AI in Public Sector

Artificial Intelligence (AI) has the potential to transform the public sector from improving healthcare and education to enhancing public safety and transportation. The growing interest in generative AI is clear. The availability of Azure OpenAI Service as a commercial offering has served as an accelerator to the consideration of public sector AI use cases.

However, with this ‘great power comes great responsibility’, and it is essential that AI is developed and deployed responsibly. Microsoft has taken a principled role in this area with the development of comprehensive AI responsibility policies and tools.

The responsible use of AI is, of course, a topic which public sector organizations around the world have actively addressed in recent years. Through leading discussions, developing approaches and strategies, and implementing these in their operations, the use of AI to responsibly deliver better and more inclusive public services is on the rise.

At Microsoft, we are committed to making sure AI systems are developed responsibly and in a way that is worthy of people’s trust. We drive this commitment according to six key principles which align closely with public sector priorities:

• Fairness: AI systems should be designed to treat all individuals fairly, without bias or discrimination.
• Reliability and safety: AI systems should be reliable and safe, with built-in mechanisms to prevent errors and minimize harm.
• Accountability: The creators of AI tools and the developers who leverage them should be accountable for their systems.
• Privacy and security: AI systems should respect individuals’ privacy and data security.
• Inclusiveness: AI systems should be designed to be accessible and usable by everyone, including individuals with disabilities.
• Transparency: AI systems should be transparent and explainable, with clear documentation of their functionality and decision-making processes.

These principles can be used by public sector organizations to evaluate AI systems and processes in use or under consideration. Within Microsoft, this responsibility is propagated by a centralized Office of Responsible AI, which sets AI governance policies for the entire company, advises our senior leadership team on AI issues, enables engineering and compliance teams across the company to build according to responsible AI principles, and ensures that as a corporation we are continuing to examine and improve our ethical stance as new capabilities and challenges arise. In June 2022, we published our internal Microsoft Responsible AI Standards for product development to share what we’ve learned so far in the form of concrete and actionable guidelines. We believe that industry, academia, civil society, and government need to collaborate to advance the state-of-the-art and learn from one another.  

Public sector organizations should develop and be governed by responsible AI strategies, and these strategies should incorporate principles, practices, tools, and governance to enable those across the organization to assess, adopt, and manage AI. 

This may involve:

• Developing new or adopt existing policies and guidelines.
• Providing training to staff to ensure they are aware of the considerations associated with AI.
• Ensuring data used to train AI models is representative and assessed for bias.
• Establish governance bodies to subject sensitive use cases to particularly high scrutiny and use tooling and telemetry to ensure that they are functioning as intended and not causing unintended harm.
• Ensuring accountability for both the development and operations of AI capabilities and AI-enabled systems.

Staff, stakeholders, and citizens can provide feedback that helps to improve individual AI workloads and the respective responsible AI strategy.

Once potential risks are well understood and carefully managed, the public sector can realize the promise of AI. Forward-looking leaders will ensure that their commitment to responsible AI is not an afterthought but is baked into their organization’s innovation pipeline. This allows the public sector to harness the power of AI in a way that improves the services provided and benefits society as a whole. 

AI Public Sector Vision and Disruption

There are moments in the evolution of technology that impact society and the economy deeply – disruptive advancements in innovation whose effects are not fully felt until years after they debut. We have seen industries changed irreversibly by the internet, smartphones, and cloud, just to name a few. Although it is hardly a new technology topic, recent advancements have catapulted AI into the firmament of game changing, disruptive technologies, and it is worth taking a step back to contemplate what the long-term implications for the public sector may be in years to come.

At the simplest level, generative AI will be a massive force multiplier for overworked government staff. Imagine social workers who can now be in contact far more frequently with their caseload using a mix of human interactions and those augmented by virtual assistants. Imagine parents and children able to access academic support and tutoring at the frequency and intensity they require without obstacles around identifying, affording, or accessing the providers of that support. Especially in the rapidly graying societies of the developed world, imagine aging populations who can interact with AI assistants designed to keep them active and engaged, optimized to combat the harmful effects of isolation. Public sector call center employees will be free to do case work once a generative AI interface can handle transactional front-line calls of all varieties. The transformative possibilities are varied and exciting.

Good government runs on high-quality interaction. Governments hire and train the best public workforce they can find to deliver excellent services, and that will never change. Yet, public sector budgets are perpetually under pressure, and the limited size of the public sector workforce is an inherent cap on the breadth and depth of services government can provide, and the speed at which they can respond to citizens’ needs. AI allows governments to radically increase capacity without dramatic new expenditures. These AI assistants will never replace their public servant practitioners, but can give them a copilot, augmenting their efforts at low cost and with high impact.

The broad span of government also makes it inherently complex to navigate, especially for citizens who only interact with public services on a sporadic basis. AI can help citizens and businesses understand and plot a course through the dense thicket of government laws and regulations, and promises to create predictive, anticipatory government services, proactively reaching out to potential beneficiaries, and identifying immediate opportunities for constituents having vaulted the bureaucratic hurdles in advance. This is a complete inversion of our delivery model for public good, with benefits automatically notifying potential recipients, instead of forcing citizens to become bureaucratic experts to even learn about the services to which they are entitled.

Another notable long-term effect for the public sector is the leap forward in accessibility AI provides. Integrated AI universal translators provide access to information independent of language preference, allowing populations all over the world to access a vast sea of knowledge in their mother tongue. But this is just the beginning. Already the public sector is experimenting with using natural language – written or spoken – to access government information and citizen services, with image and voice capabilities following quickly behind. Although the tech industry tries hard to make interfaces flexible, intuitive, and operational across a variety of devices, legacy interfaces by their nature still become a blocker to some members of society accessing needed services. Rapid, universal, knowledgeable, life-like, helpful conversation replaces interfaces. Natural interaction completely disintermediates the technology and becomes a holistic means for accessible government capabilities. For societies, language barriers become an obsolete notion, and for the public sector, a transformative capability emerges.

An additional important dynamic is the gravity these AI discoveries create between the public sector and the private industry. Public sector organizations are accustomed to keeping industry at arm’s length, pursuing separate infrastructures with dedicated instances and segregated data. However, AI thrives on broad diversity, and large language models specifically are hampered by segregation. Fencing a training set within a municipal, provincial, national, or regional boundary undermines the signature features of the model, significantly restricting its core functionality, and delivering a weaker model. This is compounded by the challenges inherent in replicating the enormous resource backbone required for training these models. Finding a way to leverage shared infrastructure for these models becomes a social imperative, as private replications of these efforts may represent an irresponsible devotion of financial and energy resources. These dynamics create a strong incentive for the private sector to provide AI models in a way that the public sector can consume compliantly, and an equally strong incentive for the public sector to find compliant ways to consume them. Democratizing access to advanced AI capabilities is the next step in Microsoft’s continuing journey to empower every person on the planet to achieve more.

AI technologies have already evolved to heretofore unseen maturity and prevalence, and these are early days. Bing’s chat interface for search, Viva Sales suggested responses, GitHub Copilot’s ease of code generation, Teams’ automated meeting summaries, and Microsoft 365 Copilot’s varied array of productivity accelerators all have broad applicability both inside and outside the public sector today. But our imagination should not be limited by these early paces in the AI marathon. It behooves us now to dream about the long-term transformative effect the AI revolution will have on the way government interacts with its citizens, and to start to build those dreams for the benefit of society.

To find out more:

• Follow Angie Heise on LinkedIn
• Insights into Generative AI Technology
• Learn more about Microsoft’s approach to Responsible AI
• Learn more about Microsoft 365 co-pilot