Transforming the public sector

The implications, risks, and best practices of implementing AI

Philipp Roth
Lead Partner, Government & Public Sector, PwC Switzerland

Philipp Rosenauer
Partner Legal, PwC Switzerland

Artificial Intelligence (AI) is poised to revolutionise many facets of public life, from healthcare and education to transportation and defence. As the public sector moves towards digital transformation, it is essential to evaluate how AI can help improve operational efficiency, citizen experience and decision-making processes, while also understanding and mitigating the associated risks. From streamlining administrative processes to enhancing citizen experiences, AI offers promising opportunities for the public sector.

Selected AI use cases in the public sector

Healthcare services:
AI can assist in early disease detection, personalised treatment plans and predictive health outcomes. Machine learning algorithms can analyse large volumes of health data to identify patterns that humans may overlook, leading to more accurate diagnoses and better patient care. For instance, Google’s DeepMind can detect over 50 eye diseases as effectively as a human specialist. This represents a significant stride towards earlier and more accurate disease detection and treatment.

Education:
AI-powered platforms can create personalised learning experiences, tracking student performance and adapting educational content based on individual needs. It also helps in automating administrative tasks, reducing workload on educators and allowing them to focus more on instruction. Georgia State University in the United States uses an AI-powered chatbot named ‘Pounce’, which helps students navigate administrative tasks, leading to a 3.3% increase in enrolment.

Transportation:
AI is transforming public transportation systems through smart traffic management, autonomous vehicles and predictive maintenance of infrastructure. By predicting traffic and transit patterns, it can optimise routes and reduce travel times.

Administrative efficiency:
AI can automate routine tasks such as document verification and data entry, thereby improving operational efficiency. It also assists in decision-making through predictive analytics and data visualisation tools. Estonia, a global leader in e-governance, leverages AI for various administrative tasks, from chatbot-based customer service to predictive analytics in policymaking.

«The era of AI in the public sector is here and its potential to improve public services is vast»

Philipp Roth,Partner, PwC Switzerland

Generative Artificial Intelligence

Artificial Intelligence is a complex and hot topic. Do you have any questions? We are here to help you.

Talk to our experts!

Unique challenges for public sector AI implementation

The increasing use of AI in public services is not without challenges. Some risks include:

Data privacy: Public institutions handle large volumes of (sometimes sensitive) personal data about their citizens. Ensuring this data is securely collected, stored and processed is paramount in order to maintain public trust. With increasing frequency of data breaches, it is crucial to ensure robust data protection measures.

Bias: AI systems can unintentionally perpetuate and amplify existing biases, leading to unfair outcomes. It is crucial to continually review and adjust AI algorithms to ensure fairness.

Transparency: Citizens have the right to know how decisions impacting them are made. AI models can be opaque and complex, making it difficult to explain how they derive their conclusions.

Equity of access: Public services must be available to all citizens. With the implementation of AI, care must be taken to ensure that these services remain accessible to those with limited digital literacy or access.

Security concerns: Generative models could be used to generate malicious content (e.g. fake news, deepfakes), resulting in reputational damage for companies.

Explainability: Clear and understandable explanations might be lacking, resulting in a lack of interpretability and accountability. This can be a challenge for decision-making processes and may raise questions regarding accountability.

To mitigate these risks, public authorities can develop a framework such as Responsible Artificial Intelligence (RAI), developed within PwC. This framework would implement various responsible practices, such as:

Adopt a privacy-first approach: Use encryption and anonymisation techniques to secure data. Regularly review and update privacy policies in line with current laws and regulations.

Invest in bias mitigation: Engage diverse teams in AI development, use representative training data and continually monitor AI systems for bias.

Ensure transparency: Opt for interpretable AI models where possible and work towards ‘Explainable AI’, ensuring outcomes can be understood by non-experts.

Ensure security: Enhance the cybersecurity of systems.

Be environmentally conscious: Minimise negative environmental impact by promoting sustainability.

Protect privacy: Develop systems that preserve data privacy.

Invest in robustness: Enable high-performing and reliable systems.

Practical guidelines for implementing AI in the public sector

Successfully implementing AI in public services requires a strategic and thoughtful approach. Here are some practical tips for public authorities:

Define clear objectives:
Before implementing AI, identify the specific challenges you wish to address and set measurable goals. This helps align AI initiatives with broader organisational objectives.

Assemble a diverse team:
Diversity in AI development teams ensures different perspectives and can help in identifying and minimising bias in AI algorithms. This includes diversity in terms of gender, race, cultural background, professional experience and more.

Invest in infrastructure and skills:
Implementation of AI requires robust digital infrastructure and skilled personnel. Investment in both hardware and training is crucial to ensure the seamless integration of AI technologies. Prioritise building a team that understands AI, its potentials and its pitfalls.

Keep citizens at the heart:
Never lose sight of the fact that the ultimate goal of AI in the public sector is to serve citizens better. Regularly engage with citizens to understand their needs and concerns, seek feedback on AI initiatives and adjust your approach based on their input.

Prioritise equity and inclusion:
As AI is integrated into public services, public authorities must ensure that these services remain accessible and beneficial for all citizens. This could include providing digital literacy training, ensuring AI services are available in multiple languages and using universal design principles.

Ethical considerations:
The implementation of AI should be guided by ethical considerations. This includes respecting privacy, ensuring fairness, being transparent about AI use and maintaining accountability for decisions made by AI systems.

Pilot and scale:
Before fully integrating an AI system into operations, conduct pilot tests to identify any potential issues or areas of improvement. Once the technology has been refined and proven successful, it can be scaled up.

Prepare for change management:
The introduction of AI may involve significant changes to work processes, job roles and even organisational structure. Be prepared for change management and ensure open communication channels to address concerns and resistance that may arise.

Regulatory compliance:
Stay updated on AI-related regulations and ensure your AI initiatives are compliant. This not only avoids legal issues but also enhances public trust in your use of AI.

The era of AI in the public sector is here. Its potential to improve public services is vast, and with careful planning and strategic implementation, it can be harnessed effectively. However, as with any powerful tool, the key lies in understanding its use, addressing the risks, and guiding its application ethically and responsibly. It’s about using AI to create a more efficient, responsive and citizen-centric public sector while ensuring privacy, fairness, transparency and accountability.

#social#