The ‘Human in the Loop’ Won’t Last Forever
Karina Wellendorph, Anja Monrad and Katarina Gospic, panelists at a Board Summit session on human judgment in AI, on why boards shouldn’t assume humans stay in the loop forever, and where judgment matters most once AI takes over the routine calls.
At a Board Summit panel on human judgment in AI — moderated by Barbara Myhre Isaksen, with panelists Karina Wellendorph, Anja Monrad and Katarina Gospic — one warning cut through the rest: the reassuring idea that humans always stay “in the loop” on AI decisions may not hold much longer. “I just think we need to be careful that we don’t think that the human in the loop thing is going to stay forever,” one panelist said — a caution from someone who described themselves as a technology optimist, not a doomsayer.
Reserve humans for what matters most
Panelist Katarina Gospic walked through a real example: automating invoice processing end to end, where AI extracts the data and rules route small transactions straight through, while only high-value deals — a million-kroner invoice, say — get a person in the loop. The goal wasn’t fewer humans. It was freeing them up to spend time on the deals that actually move the business.
Trust the process, not the tool
Karina Wellendorph, whose company handles the books for more than 300,000 Danish businesses, argued that accountability can’t be automated away. “You can never blame AI for a fault in your account,” one panelist said, echoing an old IBM line about computers. Trust in AI, Gospic added, comes from training it on your own company’s data with people you already trust — not from trusting a chatbot.
Boards are behind, and it’s not really about AI
The panel agreed boards don’t need a dedicated “AI person” any more than they needed a standalone ESG or digital specialist — several noted that many boards haven’t even reached basic digital literacy yet. What’s needed instead is a shared baseline of understanding across the whole board, because an AI strategy is really just a business strategy that happens to involve AI.
The takeaway
Judgment doesn’t disappear as decisions get automated — it moves. As routine calls get handed to AI, human attention should shift toward outcomes, values, and the harder decisions where getting it wrong actually costs something.
