Public sector data-driven decision-making enabled by AI

By Kwaku APPIAH-ADU (PhD) & Derrydean DADZIE

The digital economy is an exciting topic, and there is no doubt about it. Even more so is the glut of technological advancements that we are introduced to daily, all of them promising a vastly different world than the one we live in today. Artificial Intelligence (AI) and Large Language Models (LLMs) are the champions of the technology field today.

A few days ago, Nvidia (developer of the Graphical Processing Units, GPUs, on which AI applications are built) surpassed Apple as the 2^nd most valuable company in the world, primarily driven by their central role in the AI space. It is abundantly clear that artificial intelligence (AI) is not just the ‘next big thing’; it’s a brand-new foundation upon which today’s digital economies are being built.

However, all the Foundational Blocks of the digital economy that are spoken of generally (and in the book authored by Appiah-Adu and Asare (2022) titled “The Enabling Architecture for a Digital Economy”) are of little use to ordinary citizens who primarily experience a ‘digital economy’ via the applications made available to them. This also holds true for government officials as well, and perhaps even more so, as their work directly touches millions of lives.

If, as we say, ‘charity begins at home’, it is even more important that governments pushing digital applications in a digital economy must live in that digital space themselves.

Before we proceed, let us step back and provide some simple explanations for terms we will use extensively in this piece: Artificial Intelligence (AI), Generative AI (or GenAI), and Large Language Models (LLMs).

• Artificial intelligence (AI) is the ability of a constructed machine, such as a computer, to simulate or duplicate human cognitive tasks. A machine with AI can make calculations, analyse data to create predictions, identify various types of signs and symbols, converse with humans, and help execute tasks without manual input (per Cloudflare).
• Large Language Models (LLMs) are a type of AI that are designed to understand and generate human language. These models are trained on vast amounts of text data and can then generate human-like text and responses.
• Generative AI, or GenAI, refers to AI systems that have the ability to create new content, whether it be text, images, or other forms of media, based on the data they have been trained on. Most people are interfacing with AI via a GenAI program such as ChatGPT or Gemini.

It is worth noting that Governments worldwide are already starting to harness the power of AI and LLMs to improve governance and better serve their citizens. From predictive analytics to personalised services, AI is transforming the way governments operate.

For example, Seoul Talk in South Korea exemplifies an innovative initiative utilising AI consultants to manage citizen inquiries and complaints about city functions and services. This platform addresses a fundamental responsibility of government: informing citizens on how to access public services.

The Automatic Social Energy Tariff initiative in Portugal utilises secure, integrated data from government agencies and energy companies to identify eligible citizens for social energy tariffs and automatically enrol them in the program.

This system streamlines transactional interactions between citizens and the government, such as tax collection and benefit distribution, which are typically managed by multiple agencies at both local and national levels. Moreover, in India, the integration of voice-payment technology into the instant payments system is set to enhance financial inclusion, especially by providing support in multiple languages.

Similarly, African governments are being served with an excellent opportunity to leverage AI tools to improve service delivery for their citizens in several key areas. Broadly, here are the top three main benefits to consider:

4. One key benefit of AI in governance is its ability to improve decision-making. By analysing vast amounts of data, AI systems can provide valuable insights that human decision-makers might overlook. This can lead to more efficient policies, better resource allocation, and ultimately, improved citizen outcomes. With AI, Policy makers can design and implement interventions in complex systems, conduct real-time policy evaluations, and gain detailed insights into public sentiments.
5. Furthermore, AI can help governments provide more personalised services to their citizens. By analysing individual data and preferences, AI systems can tailor services to meet the specific needs of each citizen. This can lead to greater satisfaction among citizens and increased trust in government institutions.
6. Furthermore, AI can enable governments to deliver more tailored services to various communities. By analysing segmented data and group preferences, AI systems can customise services to address the specific needs of different communities. This approach can lead to greater overall satisfaction and increased trust in government institutions.
7. Moreover, AI systems can play a pivotal role in eliminating backlogs and reducing waiting periods within governmental operations. This technological advancement promises enhanced efficiency, diminished bureaucratic delays, elevated service quality to bolster citizen satisfaction, cost savings, and heightened public trust in government institutions.

These broad benefits outlined above can also be applied to create valuable solutions in specific sectors. Let us examine a few:

1. Healthcare: Generative AI can be used to analyse and predict disease outbreaks, such as COVID-19, Malaria, or Ebola. By using AI to process and interpret data from various sources, including health centres and hospitals, government officials can make data-driven decisions to allocate resources, implement preventive measures, and improve public health policies.

Imagine a public official being able to ask a GenAI tool fed with public health data ‘what are the communities in Ghana with the highest maternal mortality rate per district hospital data?’ in order to plan government subventions targeting specific areas instead of a more expensive national program. The speed with which tooled LLMs could make such information accessible would in itself be a game changer.

2. Agriculture: AI can significantly improve agricultural productivity and food security in the region. Generative AI can help farmers and government officials predict weather patterns, optimise crop yields, and detect pests or diseases early. By analysing satellite imagery, weather data, and soil samples, AI can provide actionable insights for policymakers.

For instance, AI tools fed with local data from the Ministry of Agriculture can be used to optimise extension service interventions for local farmers as well as priority investment areas for the government looking to protect and grow crop yields nationally.

3. Infrastructure and Urban Planning: AI can help governments optimise infrastructure development and urban planning. Generative AI can analyse data on population growth, traffic patterns, and resource usage to help officials make decisions about where to build new roads, housing, or public facilities.

For example, imagine our officials using AI to pre-determine the growth areas of Takoradi to understand the best new roadworks to prioritise to reduce traffic, beyond that to plan public facilities such as hospitals and clinics to be sure they are available within minutes of population centres.

Despite all this good news, using AI in governance also comes with challenges. Three key concerns public officials looking to integrate AI into their workflows need to consider are:

1. Sensitive Data Exposure: Government data often includes sensitive information about citizens. Using AI tools requires robust measures to protect this data from unauthorised access and breaches. It is extremely important to ensure that critical national data does not fall into the wrong hands, this can lead both to direct harm done to the nation or its citizens and, beyond that, a loss of public faith in the use of technology tools in governing.
2. Data Quality and Availability: Government data may be siloed, incomplete, or outdated. This can lead to inaccurate or unreliable outputs from AI models. A massive data clean-up and organisation exercise would be needed prior to deploying AI tools to ensure quality data is gathered and in sufficient quantities to properly and holistically train AI models.
3. Technical Expertise: A major one for African governments is that building and maintaining AI infrastructure and applications requires specialised skills that governments may not readily have. Partnering AI experts may be necessary; however, governments need to ensure that such partnership deals prioritise the training and upskilling of local expertise to ensure that within a short time frame, a country’s nationals can entirely run these platforms on behalf of their governments.

As governments continue integrating AI into their operations, they must do so transparently and ethically. Public officials need to constantly educate and bring the public they serve with them in their journey to lay this critical foundation of a modern digital economy.

How, then should public officials get started?

1. Data Infrastructure Enhancement: Invest in robust data collection and storage infrastructure. The “Enabling Architecture for a Digital Economy” authored by Appiah-Adu and Asare (2022) goes into depth on the topic of ‘Storage’ as a key foundational deliverable for a digital economy, and that holds true here as well.
2. Capacity Building: Train government officials and staff on AI technologies and their applications. Digital skills need to be prioritised at the highest levels and start at the earliest levels of the educational pyramid to ensure citizens and public officials are abreast of the latest tools and strategies.
3. Collaborations: Partner technology companies and research institutions to develop and deploy AI solutions. Governments do not need to build everything from scratch from Day 1. However, due to the sensitivity of public data, it is important to be highly circumspect in drafting any such collaborations with external parties for AI use.
4. Pilot Projects: Start with pilot projects in critical areas to demonstrate the benefits of AI. There is no need to immediately launch into projects worth 100s of billions. AI-focused digital transformation should be gradual, focusing on specific districts or government departments and agencies first to ensure a limited scope that would make it easier to measure success while simultaneously limiting risk.
5. Policy Framework: Develop a policy framework to govern the use of AI, ensuring ethical and responsible deployment. This needs to be done to provide clear guidelines across all government bodies for deploying and using AI tools and services. Beyond that, it is imperative to lay out clear goals for AI use so that the mere deployment of the software is not considered the measure of success.

AI and LLMs have the potential to revolutionise governance and improve the lives of citizens around the world. By harnessing the power of these technologies, governments can make more informed decisions, provide better services, and, ultimately, create a more efficient and responsive governance system.

Dr Appiah-Adu is the Senior Policy Advisor, and Derrydean is the Digital Transformation and Innovation Policy Advisor, CEO, Heritors Labs Limited

The post Public sector data-driven decision-making enabled by AI appeared first on The Business & Financial Times.

Read Full Story