I am Ryuto, a PhD candidate at the Institute of Science Tokyo (Formerly Tokyo Tech), advised by Prof. Naoaki Okazaki. Currently, I am visiting the UPenn NLP, hosted by Prof. Chris Callison-Burch. I also work with Prof. Preslav Nakov from MBZUAI NLP. In addition to my PhD research, I am involved as an advisor for a startup on multi-lingual text generation.
I am interested in the discrepancy between human writing and LLM generation, aiming to understand and enhance LLM generative capabilities. To explore this, I have mainly been working on two key aspects:
- 🔎 Detecting LLM-generated texts
- Robustness. How can we improve the robustness of detectors in the wild? [AAAI'24]
- Interpretability. How can we offer detection results to lay users more reliably? [In submission], How well can people detect LLM texts? What are the clues? [In submission]
- ✍️ Harnessing LLMs for better generation
- Prompting. How can we instruct LLMs to unlock their generative capabilities, narrowing the gap between human writing and LLM generation? [EMNLP'24 Findings]
- Evaluation. How can we make LLM-as-a-judge more reliable? [ACL'24 Findings]
📢 Actively looking for research internships starting in (Summer | Fall | Winter) 2025.
- Personal Website: sites.google.com/view/ryutokoike/
- Twitter: @sponddd
- email: my_first_name.my_last_name[at]nlp.c.titech.ac.jp