Keeping humans in the loop: Ingvild Bode on AI, war and the choices we still have

Ingvild Bode did not set out to become a scholar of international politics.

When she left school, she imagined becoming a teacher. Politics was not particularly high on her list of interests. Languages were. Different cultures were. Then, during an Erasmus year in Dublin, something changed.

Some of her new friends were studying politics, and Bode began joining their lectures.

Unofficially.

– I just randomly attended lectures that they also attended. I kind of snuck in at the back, they were big lecture halls, she recalls.

What began as curiosity became something more serious. When she returned from Dublin, she changed her course of study.

Politics, in a sense, had found her.

Today, Bode is Professor of International Politics at the University of Southern Denmark, Director of the Center for War Studies and a new Chair at the Danish Institute for Advanced Study. Her research sits at an intersection that has become increasingly difficult to ignore: artificial intelligence, warfare, human decision-making and the international norms governing the use of force.

Her route there was gradual. Her PhD focused on policy and normative change within the United Nations system. Later, work on armed drones brought technology into the picture during the Obama era, when their growing role in American counter-terrorism operations raised disputes about what this new way of using force meant. For Bode, the most interesting questions were normative.

– What happens to our approach to the use of force when we have new technologies that enable different kinds of practices? And how do those practices then normalize certain ways of using them? she asks.

From drones, her research moved towards autonomous weapons systems and the growing integration of AI into military technologies. At the time, mentioning autonomous weapons often produced a predictable association: Terminator. Killer robots. Science fiction.

But the reality was less spectacular and more consequential. Militaries had experimented with automation and autonomy for decades. AI was not arriving on the battlefield from nowhere.

Beyond the AI hype

Few technologies currently carry as much narrative baggage as artificial intelligence.

AI systems are frequently presented as faster, more objective and more precise than humans. In military contexts, advocates also argue that they might make warfare safer, perhaps even improving the protection of civilians.

Bode is interested in these promises, but equally in what happens when they meet reality.

– I think AI is one of these technologies that is surrounded by so many narratives. There are these big claims about how it is supposedly more objective and more precise, she says.

Shortly before our conversation, Bode had returned from a United Nations meeting in Geneva where states discussed AI in the military domain. There, too, she encountered arguments about its potential humanitarian benefits. The potential may exist, she stresses. But potential is not the same as practice.

Looking at contemporary conflicts, including the use of AI-assisted decision-support systems in Gaza, reveals a more complicated picture. Technologies presented as tools for precision exist alongside devastating civilian consequences.

For Bode, responsible AI therefore cannot simply be added afterwards as a safeguard. Decisions about legality, ethics and human control need to be considered while systems are being designed, programmed and acquired - long before they reach an active battlefield.

That also challenges another powerful story about AI, that its growing role is somehow inevitable.

Bode is skeptical of that framing. It risks presenting politicians as passive spectators to technological development.

– They can make choices. They are not just reacting to whatever, for example, the technology companies are proposing, she says.

Governments decide which systems to acquire, what standards they should meet and how they may be used.

The future is not simply something technology does to us.

What happens to the human?

At the centre of Bode’s recent research is a deceptively simple question: what happens to human decision-making when more of the cognitive work is delegated to machines?

Her six-year ERC-funded AutoNorms project has examined how states develop and use increasingly autonomous weapons systems, tracing practice rather than relying on promises. Her work continues through projects including HuMach, funded by the Independent Research Fund Denmark.

What concerns her is not one dramatic moment in which a machine suddenly replaces a human. It is the cumulative shift.

As systems identify information, classify objects, generate recommendations and support decisions, humans may gradually become more dependent on their outputs.

In warfare, the human role carries several functions at once. There is a legal one: humans remain essential to accountability and compliance with international humanitarian law. There is a safety function: AI systems are fallible. And then there is the moral question.

War confronts people with profound ethical dilemmas. Bode does not romanticize human decision-makers. Humans make mistakes. They can behave terribly.

But our concepts of responsibility, justice and ethics remain built around them.

– We need humans to remain active ethical and moral decision-makers, she says.

What happens when those people become increasingly dependent on machine-generated outputs is therefore not just a technical question. It is a question about what kind of decision-making we want.

There is rarely a ‘first AI war’

Studying that question is made harder by the speed of technological and political change.

Headlines regularly proclaim another breakthrough in AI-enabled warfare. Conflicts are described as the first “AI war”. Then another conflict earns the same title.

For Bode, good research requires resisting the temptation to treat every development as unprecedented.

– In order to understand where things might be going, you have to understand where things have come from.

Traditional academic publishing alone cannot provide that perspective quickly enough. By the time a peer-reviewed article appears, technologies and practices may already have moved considerably. Bode therefore also relies on international expert networks, closed-door discussions and “grey literature” from think tanks and specialized organizations.

Ukraine offers a useful example of why simple narratives are rarely enough.

The war is often presented as evidence of a march towards ever greater autonomy. Yet Ukrainian forces have also made extensive use of fiber-optic drones, physically tethered by extremely thin cables. One advantage is that they are difficult to jam, but they also allow operators to retain direct contact and control.

Even amid rapid automation, the human has not simply disappeared.

Crossing the boundaries

That complexity is part of what brought Bode to DIAS.

Her career has taken her through research environments in Germany, Japan, the United Kingdom and Denmark. Moving between academic cultures taught her something valuable, distance from what feels normal.

– It allows you to question what you think was normal.

As Bode went deeper into the societal consequences of military AI, the limitations of approaching the subject through international politics alone became increasingly apparent.

Understanding AI requires engaging with the technology itself. But it also raises questions of law, ethics, history, politics and culture.

– That is where I really feel DIAS is the perfect place. It allows you to entertain all of these conversations, she says.

At this point in her career, she sees room to experiment, take intellectual risks and build more interdisciplinary networks locally at SDU - including with researchers who might initially believe her work has little to do with theirs.

Military AI is Bode’s empirical world. But its central questions travel much further.

Healthcare, elderly care, public administration and education are all confronting related choices about automation, judgement and responsibility.

– How does using AI change human decision-making? she asks.

And, perhaps more fundamentally:

– What do particular ways of designing and using these technologies do to us as humans - individually, but also collectively?