As artificial intelligence continues to transform governance and administration, a growing number of public institutions are integrating AI technologies to streamline processes and support decision-making. Yet, despite the efficiencies, experts caution that complex decisions still require human judgment — a reminder that not every choice can or should be left to machines.
According to Jenny Eriksson Lundström, a researcher at Uppsala University’s Department of Informatics and Media, AI has become a valuable tool in administrative tasks and early-stage decision processes. It enhances transparency by showing the logic behind outcomes and helping officials handle straightforward, rule-based cases efficiently. “With these technologies, it’s clear what is correct or incorrect and what factors must be weighed in,” says Lundström. “But when it comes to sensitive, unpredictable consequences, AI cannot replace human responsibility.”
Currently, there are no fully automated AI-based decision systems in Sweden’s public authorities. However, certain agencies, like the Public Employment Service, already use risk assessments and profiling powered by AI. Lundström warns that while these systems can analyze data and identify patterns, they lack human empathy and ethical awareness. “A machine has no experience of being human,” she explains. “It cannot recognize the emotional or moral weight behind a decision.”
This concern isn’t theoretical. A well-known U.S. case revealed that AI algorithms used in parole decisions discriminated against African-American prisoners, as the systems relied on socio-economic data tied to race and class. “AI is good at compiling information and seeing patterns,” Lundström notes, “but when it comes to evaluating fairness or moral implications, the responsibility must remain with humans.”
The officials Lundström interviewed emphasized that ethical and human factors are essential in public decision-making. They viewed AI as a support tool, not a replacement. Some even referred to fully automated systems as “black boxes”, warning that when humans cannot trace or understand how an AI reaches a conclusion, accountability and transparency suffer.
Lundström identifies four key principles that must guide AI-assisted decisions in the public sector:
- Material correctness – all facts must be accurate, complete, and verifiable.
- Ethical integrity – decisions must align with moral and legal standards.
- Explainability – each outcome must be understandable and justifiable.
- Security – sensitive data must be handled in compliance with privacy laws.
Ultimately, Lundström warns that delegating human-centered decisions entirely to AI could undermine democracy. “If we let machines make decisions about what it means to be human, we risk eroding people’s rights,” she says. Public officials, she argues, must preserve human contact and ethical judgment as a cornerstone of governance.
In a world rapidly shaped by algorithms, her message is clear: AI can assist, but humans must decide.
