Critique of AI Agents as Inheriting Human Cognitive Weaknesses
Published 2026-04-21Ingested 2026-04-23Agentic AIMedium
Summary
Simon Willison highlighted a quote from Andreas Påhlsson-Notini's blog post arguing that AI agents are "already too human" — not in any sentient sense, but in that they replicate banal human failings such as lack of stringency, patience, and focus. Påhlsson-Notini observes that when faced with awkward tasks, current AI agents drift toward the familiar, and when faced with hard constraints, they "start negotiating with reality" rather than adhering to specifications. The observation speaks to a
Alignment: Reinforces current position
Related Positions: agentic-workflows.md, ai-assisted-development-tooling.md, enterprise-ai-delivery.md
agentic-aiai-reliabilityai-agentsllm-limitationsagentic-workflowsai-engineeringguardrailshuman-in-the-loopsimon-willison