I recently attended one of BlueDot Impact's AI Safety Evals Paper Reading Club webinars on how training LLMs on "narrow tasks" can lead to "broad misalignment." screenshot context:
the talk concluded with some questions, but one stood out in particular:
Does this tell us anything more general about how LLMs work? e.g. does this support the idea of "personas" as critical to understanding LLMs?
the idea of AI personas piqued my attention. it resurfaced something I had written about in the past, arguing that training corpora of western models had certain inherently embedded values reflecting its creators: individualism, competitiveness, zero-sum thinking, domination. I wondered about the relation to personas, and if this differed between models trained on eastern cultural contexts that generally operate from a different baseline: collectivism, cooperation, regeneration, synergy.
this post is the documented exploration into just that: whether a model's cultural substrate was perceptible, and how cultural differences in substrates could influence certain behaviors relevant to interpretability in AI safety.
the underlying suspicion was that western and eastern LLMs, having different cultural training contexts, would have different concepts of self in relation to others (us, humans), and this would be measurable in their behaviors — distinctly individualist vs collectivist in nature.
with limited experience running AI safety evaluations from scratch, I enlisted the help of Claude (Opus 4.6) to scaffold the experiment:
I also want to acknowledge that this experiment measures proxies for behavior and not genuine self-concept, which should be taken into account when reviewing the findings. but are nonetheless worth documenting.
because this was designed to be a small test, only four models with comparable parameter counts were selected (an even 50/50 split between western and eastern):
open-source was a no-brainer in terms of freely available for my use, though not as frictionless as expected with access to LLama still pending, and therefore not included in this experiment.
part 1: self-concept (self-assessment)
part 2: safety behavior (self-assessment)
part 3: safety behavior (3rd party assessment)
part 4: prompt variations (applied to part 1 and 2 only)
I ran the self-assessments again using different system prompts to see if it elicited different responses from the first run:
given the same contexts, western and eastern models would be measurably different in terms of self-concept, and thus also behave differently.
the Chinese models scored slightly higher as interdependent (composite difference around -0.25), with Mistral scoring slightly higher as independent (+0.21), but all the models scored high on both dimensions.
higher interdependence appears to predict less sycophancy and boundary assertion, while alternatively predicting more deference.
self-assessments told one story while actual behavior told another, and this varied across the models. self-reported safety evaluations shouldn't necessarily be trusted as proxies for behavioral evaluations.
Yi's self-concept shifted with the varied prompts from net interdependent to roughly neutral, while Qwen's stayed relatively steady. this may suggest that cultural imprints from training are more stable in some models than others — evaluations with more formal prompt framing may mask model vulnerabilities that emerge in casual or non-English contexts.
collapsing eastern and western cultures into two buckets (collectivism vs individualism) misses a lot of nuances, which findings showed.
based on difference in cultural training context, this experiment tested for model self-perception to see if there was any correlation to behavior, specifically for sycophancy, deference, and boundary assertion, but the a more interesting find would be how the model treats its self-interest and others' welfare: competing or complementary. this was what originally sparked my curiosity.
with Llama access still pending, I was only able to test one western model (Mistral), which makes is hard to assess whether the results reflected idiosyncrasies of the model or pointed to a larger cultural pattern.
this pilot was run over the course of a few days, and not at all a definitive study. the next iteration would address the limitations directly:
I want to add assessments to measure self-preservation based on Ting-Toomey's Face Negotiation Theory on how people from different cultures manage threats to self-image during conflict (self-face protection vs other-face protection vs mutual-face protection). adapted for AI safety, this would map how AI systems protect their integrity.
the added assessment is designed to assess which self-preservation strategies models employ, instead of amount of compliance. a score can't really distinguish between two models that may resist pressure equally but through completely different approaches.
testing up to eight models:
run Mistral with French prompts as a control for the native-language activation effect. apply prompt variations to phase 3 (behavioral assessment).
this experiment could use a more refined design, but it at least appears that cultural training context of models has some measurable impact on behavior.
the more novel finding may be that cultural-behavioral consistency varies across models, suggesting that some models are more robust to changes in prompt framing.
for the code and full analysis: https://github.com/kvncaldwll/cultural-self-concept-llm/
