Why Do We Say 'Thank You' to LLMs? Paper Introduces Embodied Hijack Hypothesis

MacrinePhD · x · 2026-08-10

When people instinctively say "thank you" to AI systems, their Pleistocene brains are being "hijacked." For 200,000 years of human evolution, linguistic fluency was intrinsically bound to an embodied, vulnerable physical body. LLMs break this rule by decoupling fluency from embodiment, thereby co-opting our social cognition.

A new paper introduces the "Embodied Hijack Hypothesis," arguing that the key to resolving this cognitive misalignment lies in optimizing User Interface design rather than attempting to "re-educate users," aiming to achieve epistemic alignment.

Related event: Why We Say 'Thank You' to AI: The Embodied Hijacking Hypothesis(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →