AI is coming to government, but civil servants and data must be ready

GREAT BAY--Governments looking to expand their use of artificial intelligence should focus as much on preparing their employees and strengthening their data systems as they do on acquiring new technology, with a lack of AI skills identified as one of the major barriers to effective adoption in the public sector.

That was one of the central messages of an analysis published by Salesforce and carried by Dutch public-sector platform iBestuur. The article argued that AI has the potential to improve government operations in areas ranging from data processing and cybersecurity to healthcare, infrastructure, fraud detection and public services, but those benefits depend heavily on the people using the technology and the quality of the information available to it.

According to the analysis, governments face a skills challenge as AI becomes more integrated into everyday operations. Public-sector employees increasingly need an understanding not only of how AI tools work, but also of data management, security, accessibility and the limitations of automated systems. Without those capabilities inside government, organizations risk adopting technology faster than their employees are prepared to manage it.

The issue has become increasingly important as governments explore AI for tasks that traditionally consume significant amounts of staff time. Potential applications identified include processing large volumes of documents, analyzing public-safety information, identifying infrastructure problems, detecting fraud, improving digital services and helping government agencies respond more efficiently to questions from the public.

The argument is not that AI should simply replace government employees. Instead, the technology could increasingly perform repetitive or lower-level tasks while employees concentrate on work requiring judgment, expertise and direct interaction with the public. That transition, however, requires governments to invest in training so workers can move alongside the technology rather than being overtaken by it.

Governments Need Their Own AI Knowledge

One of the challenges highlighted is that governments cannot depend entirely on outside technology companies for AI expertise. Public institutions handle sensitive information and operate under legal, security and public-accountability requirements that differ from those of ordinary businesses.

Employees therefore need enough knowledge to understand what AI systems are doing, where the information being used comes from, what limitations exist and when human intervention is required. Training also becomes important for ensuring that automated systems are used in ways that are secure, accessible and consistent with the responsibilities of government.

Other countries have already experimented with large-scale AI education. Finland introduced its Elements of AI initiative in 2018 to provide basic AI education to the public, with the courses eventually reaching about one million participants worldwide. Singapore has also developed AI education initiatives aimed at helping individuals and organizations understand and apply the technology.

Such programs demonstrate that AI preparation does not have to be limited to programmers or technology specialists. As artificial intelligence spreads into finance, administration, communications, planning and public services, basic AI literacy is becoming relevant across government departments.

Good AI Depends on Good Data

Training employees is only part of the equation. Salesforce also identifies data as a fundamental requirement for reliable artificial intelligence.

AI systems depend on the information available to them. Government databases that are outdated, fragmented across departments, inconsistent or poorly maintained can limit what AI systems can accomplish and increase the possibility of inaccurate results. Introducing sophisticated AI on top of weak data systems does not automatically solve those underlying problems.

Governments considering wider AI adoption therefore face a less visible task: organizing, modernizing and standardizing their information. That can include connecting databases, establishing rules governing how information is collected and stored, improving cybersecurity and determining who is responsible for the accuracy and use of government data.

The growing importance of data has also contributed to organizations establishing positions such as Chief Data Officer, with responsibility for data management and the responsible use of data-driven technology. For governments, those responsibilities can extend beyond efficiency to questions involving privacy, fairness, security and public accountability.

AI Could Change the Public's Experience With Government

The potential impact ultimately reaches beyond internal government operations. AI could change how residents interact with public agencies, particularly where services involve lengthy forms, large administrative backlogs or information spread across several departments.

Properly implemented systems could help government employees locate information faster, automate routine administrative work and provide residents with quicker access to basic services. AI could also help agencies analyze large datasets that would be difficult for employees to examine manually, potentially allowing problems to be identified earlier.

But technology alone will not improve public service. An automated system built on inaccurate information or operated by employees who do not understand its limitations could create new problems rather than solve existing ones.

That makes preparation increasingly important as governments consider where and how to introduce AI. Public institutions will need to determine which services are appropriate for automation, what decisions must remain under human control, how residents can challenge incorrect outcomes and how sensitive government information will be protected.

The larger message is that governments preparing for artificial intelligence cannot focus only on buying AI products. They must also invest in the people expected to use them and the data on which those systems will depend.

As AI becomes more capable and increasingly integrated into public administration, governments with trained employees, organized data and clear safeguards will be in a stronger position to use the technology effectively. Governments that adopt AI without building those foundations risk introducing powerful new tools into institutions that are not yet prepared to manage them.

‍