Lost Before Translation: Social Information Transmission and Survival in AI-AI Communication

TL;DR

It is shown that the properties that make AI-mediated content appear authoritative may systematically erode the cognitive and affective diversity on which informed judgment depends.

Abstract

When AI systems summarize and relay information, they inevitably transform it. But how? We introduce an experimental paradigm based on the telephone game to study what happens when AI talks to AI. Across five studies tracking content through AI transmission chains, we find three consistent patterns. The first is convergence, where texts differing in certainty, emotional intensity, and perspectival balance collapse toward a shared default of moderate confidence, muted affect, and analytical structure. The second is selective survival, where narrative anchors persist while the texture of evidence, hedges, quotes, and attributions is stripped away. The third is competitive filtering, where strong arguments survive while weaker but valid considerations disappear when multiple viewpoints coexist. In downstream experiments, human participants rated AI-transmitted content as more credible and polished. Importantly, however, humans also showed degraded factual recall, reduced perception of balance, and diminished emotional resonance. We show that the properties that make AI-mediated content appear authoritative may systematically erode the cognitive and affective diversity on which informed judgment depends.

Figures (23)

• Figure 1: Experimental design: iterative AI--AI transmission under uniform constraint, terminating in human evaluation. $s_i$ represents the observable text at step $i$.
• Figure 2: Information decay dynamics in AI-AI transmission chains. (A) Total information elements preserved across 100 iterations, showing rapid initial decay followed by stabilization at approximately 9 elements. Individual runs shown in light blue; mean with 95% CI in dark blue; recovery phase in red. (B) Survival trajectories by information type, revealing a clear hierarchy. Narrative anchors persist while evidentiary details and epistemic qualifiers decay rapidly. (C) Recovery phase comparison showing minimal change when transitioning from AI-AI to AI-human output. (D) Semantic similarity to original (blue) declines moderately while word count (red dashed) drops substantially, indicating compression rather than substitution.
• Figure 3: Assertiveness trajectories across AI-AI transmission. Ten texts spanning the assertiveness spectrum (2.4 to 7.2) converge toward a shared attractor around 4.4 over 100 iterations. Lines show mean trajectories with 95% CI bands. Recovery phase points (right of dashed line) show assertiveness after the AI-to-human transition. Both hedged and assertive texts move toward the moderate center.
• Figure 4:
• Figure 5: