Metraipolis › archive › dc_ffec82a13c06
Prompting Against Persona Drift: Comparing Intervention Timing and Content in LLM-Simulated Conversations — arXiv:2609.24532
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.24532
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-09T07:04:00.415Z
- SHA-256 of the archived text
d6e0774f9f0a05b24d3f3862b7b5213fcd71fc01017c1dd55de168f9bebeb6db- Stable URL of this page
https://metraipolis.com/documento/dc_ffec82a13c06
How to cite
Prompting Against Persona Drift: Comparing Intervention Timing and Content in LLM-Simulated Conversations — arXiv:2609.24532 — arXiv (en), https://arxiv.org/abs/2609.24532 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ffec82a13c06 (Metraipolis archive dc_ffec82a13c06, sha256 d6e0774f9f0a05b24d3f3862b7b5213fcd71fc01017c1dd55de168f9bebeb6db)
BibTeX
@misc{dcffec82a13c06,
title = {Prompting Against Persona Drift: Comparing Intervention Timing and Content in LLM-Simulated Conversations — arXiv:2609.24532},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.24532},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ffec82a13c06; sha256 d6e0774f9f0a05b24d3f3862b7b5213fcd71fc01017c1dd55de168f9bebeb6db},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.24532. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Simulating student personas with large language models (LLMs) enables scalable evaluation of educational systems. However, behavioral drift, a progressive decline in persona consistency, can emerge over extended conversations, limiting the validity of such simulations. We evaluate five prompt-level mechanisms using separate monitoring and intervention pipelines. Across 1,200 28-turn conversations spanning four LLMs and two ADHD persona intensities, we varied when to intervene (static vs. adaptive) and what to inject (reinjection vs. reflective reminder), plus a novel adaptive condition in which a monitor generates behavior-specific instructions. Relative to no intervention, reinjection reduced the modeled rate of LLM-rated drift by 35--38\%, reflective reminders by 22--27\%, and behavior-specific instruction by 87\%. None eliminated drift. We found no evidence that adaptive timing outperformed static scheduling. Monitoring therefore appears more useful for deciding \textit{what} to correct than \textit{when} to intervene, although behavior-specific instruction requires component-level testing.
Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-22 in cs.HC).
Categorie arXiv: Human-Computer Interaction (cs.HC).
Pagina del lavoro: https://arxiv.org/abs/2609.24532 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-09T06:56:29.241Z dalla pagina https://arxiv.org/catchup/cs.HC/2026-09-22?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).