What is Jiaozi: Exploring User Control of Language Style in Multilingual Conversational Agents
Bibliographic record
Abstract
Recent advances in language models have significantly expanded the capabilities of AI-powered conversational agents. Nonetheless, current technology is still primarily designed with monolingual English speakers in mind, overlooking the need of more personalized agents by multilingual users. Particularly, prior work showed that multilingual individuals preferred conversational agents that accommodate their desired multilingual style. However, these approaches rely on probabilistic methods to automatically determine the agent’s multilingual style, which often fails to align with the needs of multilingual users, as their preferences are nuanced, ad hoc, and difficult to predict. In our work, we explore user control of multilingual style as a step toward developing a mixed-initiative multilingual conversational agent tailored to the needs of multilingual users. We first derived design considerations and dimensions of user control from a formative study with 10 participants. Next, we implemented Mirrios, a prototypical conversational system with multilingual style control, and used it as a probe to conduct an user study with 12 participants. We identified preferred designs for multilingual style control and found that this control reduced the need to constrain language habits, accommodated ad hoc language needs, and enabled more personalized interactions with conversational agents. Based on our findings, we propose design implications to inform the design of multilingual style control and future mixed-initiative multilingual conversational agents.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.004 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".