Uploaded August 2026 | Updated November 2026, 1 day ago
David Krueger explains why "keep a human in the loop" won't survive competitive pressure. As agents take on longer, more open-ended tasks, the alignment bar rises — and the human supervising the work becomes a drag on productivity. The endpoint he describes is a full handover of control, where getting the values slightly wrong is catastrophic.
David Krueger is an AI safety researcher and the founder of Evitable.
#AI #AISafety #aiagents
David Krueger explains why "keep a human in the loop" won't survive competitive pressure. As agents take on longer, more open-ended tasks, the alignment bar rises — and the human supervising the work becomes a drag on productivity. The endpoint he describes is a full handover of control, where getting the values slightly wrong is catastrophic.
David Krueger is an AI safety researcher and the founder of Evitable.
#AI #AISafety #aiagents










