5 reasons for the failure of AI in public authorities

Numerous administrative processes could be optimized through artificial intelligence (AI). German studies show that up to 82% of administrative staff could be relieved by AI technologies. Nevertheless, successful AI projects in public administration are a rarity. Why is that?

While AI is already being used in many companies in the private sector to optimize processes, its application in public administration is often still in its infancy. The majority of organizations are in an experimental or exploratory phase in which fundamental questions about technology, data protection and possible applications need to be clarified.

Another problem is that many administrative processes are highly standardized and offer little room for innovative approaches. While companies can react agilely to new technologies, public authorities are often bound by stricter legal frameworks. These often slow down change. There are therefore numerous hurdles to overcome in order to bring AI into the everyday life of public administration:

Public administration is subject to strict data protection and regulatory requirements. Many AI projects fail as early as the planning phase because it is unclear how the legal requirements can be met. The uncertainty surrounding the GDPR and other regulations often leads to hesitation when it comes to introducing AI solutions.

There is often a lack of standardized guidelines or experience in handling sensitive data, which makes the implementation of AI applications even more difficult. Authorities are concerned about data protection violations, which can be associated with high penalties. There are also inconsistent requirements at federal and state level, which delays administrative projects. As a result, many authorities are reluctant to embark on AI projects in the first place.

Although AI technologies offer enormous potential, there is often a lack of specialist knowledge. IT departments are often busy operating existing systems, while knowledge of modern AI frameworks or machine learning models is lacking.

The challenge begins with selecting the right AI technology: many authorities are unsure which solution is the right one and what they need to look out for. A lack of expertise means that pilot projects often fail or are not scalable. Collaboration with external IT service providers is also difficult, as there is often a lack of internal expertise to evaluate offers and implementations in a well-founded manner.

Further training programs are rare and there is a lack of structured programs to prepare administrative staff for AI applications. Without targeted training and further education, digitalization in administration remains a slow process.

AI implementations not only require technological knowledge, but also corresponding budgets. A lack of financing options or inadequate profitability calculations mean that AI projects are not prioritized or even rejected.

Many public budgets are under pressure and investments in AI often compete with other digitalization measures or urgent administrative tasks. The costs of operating and maintaining AI systems are also frequently underestimated. In addition, public authorities often find it difficult to quantify the economic benefits of AI projects precisely, which makes decision-making even more difficult.

Many public institutions are still working with fragmented or outdated IT systems that are not designed for modern AI applications. In addition, many use cases may not be implemented in the public cloud. Strict data protection guidelines and regulatory requirements make it necessary to operate AI applications within sovereign IT infrastructures. This significantly increases the hurdle for implementation, as authorities have to rely on specialized, often more expensive solutions that meet the high security and compliance requirements. The integration of new technologies into existing processes often requires considerable adjustments, which brings many projects to a standstill.

The introduction of AI technologies is often accompanied by fears and uncertainties. Employees fear for their jobs or are skeptical about automated decision-making processes. Without an open culture of innovation, even the best technology cannot be implemented successfully.

Another hurdle is often the lack of communication about the benefits of AI. Many administrative employees have little contact with AI technologies and see them as a threat rather than a support in their work. A lack of trust in automated systems leads to employees wanting to retain their previous working methods.

Change management strategies and transparent communication about the benefits of AI applications are crucial for reducing reservations and creating acceptance. In addition, authorities must involve all relevant stakeholders at an early stage in order to create trust. This is the only way to ensure that AI is not perceived as a technological threat, but as a helpful tool.

Public administration can become more efficient, modern and citizen-friendly through the use of AI. However, the existing challenges must be actively tackled. Stronger cooperation between politics, business and research as well as the development of digital skills are essential. Only with a targeted strategy can AI be integrated into administrative modernization in the long term.