Researchers present Gricea, an open-science platform that represents conversational AI studies as configurable, deployable research artifacts others can run, inspect, share, and reuse. It addresses fragmented reporting that hinders replication in CAI research. In a replication test it reproduced 93% of eligible CUI 2026 paper configurations, and in a user study researchers from diverse backgrounds built runnable studies.
A new arXiv paper proposes using agentic AI to reimplement interactive software directly from research papers, addressing the widespread unavailability of HCI research artifacts. The authors introduce a 'revibeability' metric to measure success and demonstrate the approach on recent UIST papers.
A new arXiv paper proposes using agentic AI to reimplement interactive software directly from research papers, addressing the widespread unavailability of HCI research artifacts. The authors introduce a 'revibeability' metric to measure success and demonstrate the approach on recent UIST papers.
Researchers introduced TRAIL (Team Research and AI Integration Lab), a web platform for reproducible, configurable human-AI teaming experiments. It allows AI teammates to be designed with Big Five personality traits, selective communication, and dual memory, enabling longitudinal studies of trust and coordination.
Researchers introduced TRAIL (Team Research and AI Integration Lab), a web platform for reproducible, configurable human-AI teaming experiments. It allows AI teammates to be designed with Big Five personality traits, selective communication, and dual memory, enabling longitudinal studies of trust and coordination.
Researchers introduce CAFE (Compound-AI Factorial Evaluation), an open-source platform that applies design of experiments to evaluate compound AI systems. Practitioners can register interchangeable components (retriever, model, prompt) as factors, run factorial designs, and score answers using an LLM judge and human raters.
Researchers introduce CAFE (Compound-AI Factorial Evaluation), an open-source platform that applies design of experiments to evaluate compound AI systems. Practitioners can register interchangeable components (retriever, model, prompt) as factors, run factorial designs, and score answers using an LLM judge and human raters.
Researchers present SciForge, an AI-native workbench that integrates papers, code, datasets, and model outputs into a coherent, auditable research state. It reserves the graphical interface for human judgment while modular agent-accessible services handle search, parsing, model routing, workflow execution, plotting, writing, and presentation generation.
Researchers present SciForge, an AI-native workbench that integrates papers, code, datasets, and model outputs into a coherent, auditable research state. It reserves the graphical interface for human judgment while modular agent-accessible services handle search, parsing, model routing, workflow execution, plotting, writing, and presentation generation.
Researchers introduce Pairit, an online platform for designing, testing, and deploying live experiments on human-AI organizational designs. It addresses a gap in existing programmable platforms, which coordinate human-to-human sessions or human-AI chat but cannot declaratively specify protocols where AI participants both communicate and act on shared work. The platform aims to make such experiments auditable within a single configuration.
Researchers introduce Pairit, an online platform for designing, testing, and deploying live experiments on human-AI organizational designs. It addresses a gap in existing programmable platforms, which coordinate human-to-human sessions or human-AI chat but cannot declaratively specify protocols where AI participants both communicate and act on shared work. The platform aims to make such experiments auditable within a single configuration.