AI Paper Atlas

Research map / Large language models / AI and society

F.A.C.U.L.: Language-Based Interaction with AI Companions in Gaming

Wenya Wei, Sipeng Yang, Qixian Zhou, Ruochen Liu, Xuelei Zhang, Yifu Yuan, Yan Jiang, Yongle Luo, Hailong Wang, Tianzhou Wang, Peipei Jin, Wangtong Liu and 3 more

AAAI 2026 · Large language models › AI and society · approach: Human-centered trust and agency · 3 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Train BERT with fuzzy intent detection (entropy maximization on unsupported intents) (ablation)BERT -> BERT-FID · instruction reasoningaccuracy: 0.714 -> 0.749Improved
Fine-tune Qwen2-1.5B on domain-specific command-action dataQwen2-1.5B · instruction reasoningaccuracy: 0.642 -> 0.908 (+26.6%)Improved
Domain-aligned 1.5B LLM vs larger general Qwen2-7BLLM* · instruction reasoningaccuracy: 0.807 -> 0.908Improved
Route low-confidence commands from BERT-FID to LLM* instead of using LLM* aloneBERT-FID & LLM* · instruction reasoningaccuracy: 0.908 -> 0.886Hurt
Route low-confidence commands from BERT-FID to LLM* instead of using LLM* aloneBERT-FID & LLM* · instruction reasoningQPS: 245 -> 813Improved
Use BERT-FID instead of original BERT as the router with LLM* (ablation)BERT & LLM* · instruction reasoningaccuracy: 0.831 -> 0.886Improved
Fine-tune CLIP on in-house asset labels (ablation)CLIP · entity retrieval Top 5F1: 0.429 -> 0.790Improved
Combine text similarity with fine-tuned CLIP instead of text similarity alone (ablation)entity retrieval · Top 5F1: 0.888 -> 0.908Improved
Combine text similarity with fine-tuned CLIP instead of text similarity alone (ablation)entity retrieval · Top 20F1: 0.941 -> 0.960Improved
Full F.A.C.U.L. (BERT-FID routing, LLM*, CLIP) vs fast BERT-FID plus text similarityF.A.C.U.L. · end-to-end recognitionaccuracy: 73.9% -> 87.2%Improved
Full F.A.C.U.L. vs slow LLM* plus fine-tuned CLIPF.A.C.U.L. · end-to-end recognitionaccuracy: 89.2% -> 87.2%, QPS 210 -> 916Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy