In Germany, the hype around artificial intelligence has long since reached all levels of government. Even municipal decision makers feel the urge to look for AI-solutions in their respective areas. Newspapers and magazines for civil servants regularly report on new AI applications all over the country, from Greifswald to Berchtesgaden. LinkedIn is full of consultants and vendors pitching AI as the solution to all of Germany’s administrative digitalisation problems.
Yet we need to ask ourselves: Is AI the answer for public administrations? And to what question, exactly? Instead of focusing on a technology-driven approach that looks for any possible application of AI, we suggest that the hype money should be redirected to more pressing areas.
The AI hype wave is far from the first to hit German administrations. Does anyone still remember blockchain? The federal state of North Rhine-Westphalia introduced blockchain technology to verify the information published on its open data portal Open NRW. They were following the example of the Austrian city of Vienna, which experimented with blockchain, but did not pursue the project for long. In North Rhine-Westphalia, however, an Ethereum blockchain is used for validation to this day.
Hype directly affects funding
For a research article in the Journal for Technology Assessment in Theory and Practice, a group of scholars conducted qualitative interviews with public interest technology experts in Germany, asking about the potential and risks of AI in their sector. One finding was that the term AI was used to apply for public funding even when the projects only involved simple data analysis. This was a coping strategy to attract funding when addressing problems that did not necessarily require AI to solve.
Beyond the potential future applications of AI, the authors of the study emphasised: “limiting funding to AI technology neglects the less visible tasks and parts of a project that are often underfunded already: maintenance, capacity building, or foundational data structures”.
This key finding of the study is particularly important given the state of administrative digitisation in Germany, where these basic structures are often neglected at all levels of government. In a workshop we held at re:publica 24, we asked the obvious question: Can we not try to channel this hype money into the foundations that we need to build anyway if we want a more functional digital state? In other works, AI washing for the public good?
To our surprise, two-thirds of the participants in the workshop aimed at civil servants said they had already implemented similar strategies in their own departments. They also reported that the fixation on AI for new projects is particularly strong among senior managers, while technical staff have to explain the current limitations of the technology. Limitations such as non-existent training data, non-digitised processes in government, or possible hallucinations when dealing with large language models.
“AI” as a marketing strategy
The term “AI” is often used as a marketing strategy to create the impression that AI solves problems in a magical way, without needing to understand how. This “AI magic wand” functions as a kind of symbol to make people believe that AI takes care of everything.
It is overlooked that in reality, the use of AI often requires more competencies than conventional methods. This is also reflected in a publication by the Bertelsmann Foundation that defines a list of 21 necessary skills for the responsible application of AI technologies. Considering that a lack of data literacy among public servants is often cited as a fundamental problem for Germany’s digitalisation, adding more requirements to the wish list at this point in time seems to be counterproductive.
Part of our workshop was to create elevator pitches for fictional AI projects that would actually create a better data pool, increase data literacy, or improve knowledge and data management.
During the discussion it was pointed out that simply putting AI in the headline is not enough to get projects funded. While the two magic letters “AI” currently act as a door-opener to get the attention of senior management, the “washing” needs to be done in more subtle terms. One line of argument could be that in order to achieve the goal of a comprehensive AI application that would make all other mayors blush, the data must first be in order – machine-readable, using interoperable standards, and preferably provided automatically.
A popular example that has gained traction over the past year is using AI chatbots on websites. While this is presented as an accessible introduction to AI, the same questions as with other applications arise: Is the use of such chatbots justified in terms of a cost-benefit analysis? Do they require additional specialists and expensive licences to operate effectively? Could simpler, more durable solutions implemented internally be more efficient? Might improvements in user experience or direct website optimisations be more beneficial than a supposedly innovative AI “magic wand”?
Balancing technological and social innovation
Public administrations face the challenge of balancing technological and social innovation. Technological innovations such as AI systems promise quick results, but come with high costs and the risk of rapid obsolescence. Social innovations focus on human aspects and aim to improve societal interactions and structures. They’re often cheaper and more sustainable, but take longer to show results. AI washing appears in this context as the practice of labelling projects as AI-driven without substantial justification. It’s a response to the pressure to innovate and budget constraints in public administration.
Paradoxically, AI washing can bridge technological and social innovation. By using AI terminology in funding applications, authorities could secure resources for basic digitisation or organisational improvements, potentially supporting both types of innovation in the long term. It is a creative response to the breathlessness to succumb to anything that looks like innovation in the funding landscape.
In this way, the public sector can reap the potential benefits of AI while avoiding the pitfalls of indiscriminate adoption and misallocation of resources, investing instead in enduring foundations such as data infrastructure.
Of course, there are already AI applications that have the potential to reduce the time required to complete repetitive tasks, enhance the quality of texts, and facilitate the retrieval of internal knowledge within the German bureaucracy. But if there’s little practical benefit to be gained, the state can’t afford to use technology just for technology’s sake. In light of the current budgetary constraints and the countless vacancies at all levels of government, it is imperative to prioritise investment in sustainable, future-proof infrastructure and creating internal competencies.
However, in times of hype, focusing on such “invisible” infrastructure becomes less favourable compared to shiny solutions coming with all the buzzwords. So sometimes a thin layer of AI on top of a comprehensive data management project could be beneficial for the public good.
Dénes Jäger is a Freelance Journalist and a Project Coordinator at Open Knowledge Foundation Deutschland e.V.
Damian Paderta is a Digital Consultant who operates under the name Nozilla.
