Trang chủ / Tin tức / Khoa học & công nghệ
Khoa học & công nghệ

Steerable Chatbots: Exploring Direct Manipulation Interfaces for LLM Personalization via Activation Steering

31/10/2026 · Thành viên: Ông. Nguyễn Đức Hải · Nguồn: OpenAlex

Personalizing LLM responses typically requires users to articulate their preferences through prompting, which can be burdensome at cold start and difficult to articulate in natural language. We int...

<p><strong>Tác giả:</strong> Jessica Y. Bo, Tianyu Xu, Ishan Chatterjee, Katrina Passarella-Ward, Achin Kulshrestha, Donghee Shin</p>

<p>Personalizing LLM responses typically requires users to articulate their preferences through prompting, which can be burdensome at cold start and difficult to articulate in natural language. We introduce an alternative paradigm, steerable chatbots: rather than asking users to describe what they want, let them directly manipulate it via a linear factor. We implement this through activation steering, leveraging a linear scalar to control how strongly a preference is expressed in the LLM’s outpu...</p>

<p>📄 DOI: <a href="https://doi.org/https://doi.org/10.1145/3830398.3830533" target="_blank" rel="noopener">https://doi.org/10.1145/3830398.3830533</a></p>

<p>📖 <a href="https://doi.org/https://doi.org/10.1145/3830398.3830533" target="_blank" rel="noopener">Xem bài báo đầy đủ</a></p>

Bài viết được biên soạn lại bởi LIFVietnam. Nguồn tham khảo: OpenAlex