Across 160,860 citation-level records, the same platform surfaced systematically different source sets on its app and web interfaces.
Scale: four mainstream Chinese platforms across web and app — eight interfaces, 614 queries, three replications per query-platform-interface combination; 214,119 raw records cleaned into 160,860 citation-level records.
Three experiments on skincare: identical specifications produce near-total dominance for known brands — and less than a +0.1-star edge for a competitor breaks it.
Setup: skincare, a category where consumers cannot judge quality before buying and must rely on reputation, tested across three commercial models (GPT-4o-mini, Claude Sonnet, Gemini 3 Flash), with a robustness check on search goods.
Hold the title and abstract constant, change only author prestige, venue and citation counts, and eight models change their recommendation — while claiming they do not.
Design: authority metadata — author prestige, venue and citation counts — varied across three counterfactual conditions (original, flipped, boosted) with titles and abstracts held constant.
Ask in Chinese, Traditional Chinese or English and AI often answers differently — a documented model-level property, not a one-off glitch.
Multiple studies converge on the same phenomenon: multilingual models give inconsistent answers to semantically equivalent questions across languages, which is why cross-lingual consistency (CLC) has become its own research area.
For the same queries, Google results and generative AI answers diverge systematically in source domains, source types, query intent and freshness.
Scale: a large-scale empirical study comparing results returned by Google Search against answers from leading generative AI services.
In university applications, travel and relocation tasks, model recommendations shift with the language — local language bias is prevalent.
Scope: three personalised-advice scenarios — university applications, travel and relocation — tested across multiple languages for decision-making tasks.
Non-English queries surface fewer citations and are routed to language-appropriate sources far less often — translation alone is not a multilingual strategy.
Audit: 20 English mental-health questions under two prompt conditions, a subset translated into six further languages — 1,140 responses, 15,942 citations and 1,713 unique domains recorded.
By mid-2026, generative AI reached more than 700 million users in China with penetration above 50% — customers are asking AI which option is best.
CNNIC's Generative AI Application Development Report (2026): by the first half of 2026, China's generative AI user base passed 700 million people, with penetration above 50%.
1,909 English-only questions, six models, 30 brands: Chinese models mentioned brands in 88.9% of cases versus 58.3% for international models — driven by training-data geography, not language.
Study design: 1,909 pure-English questions across six models (GPT-4o, Claude, Gemini, Qwen3, DeepSeek, Doubao) and 30 brands.
Source structure changes by market: for Polish brands the top source is YouTube, and Lithuanian is the one language where Wikipedia isn't first.
Sample: 128 brands across 12 home markets and 13 languages — 167,551 URL-grounded citations (189,974 attribution rows).
Posting the same content everywhere averages a 10.6% citation rate; rewriting it per platform averages 38.3% — a 3.6× gap.
Zhihu is the only platform cited frequently by every Chinese engine: Kimi ~18%, DeepSeek ~15%, Doubao ~12%, Qwen ~10% — all four are engines we monitor.