Advancements in Human-AI Interaction

The field of human-AI interaction is moving towards more adaptive and interactive engagement, with a focus on generative interfaces, cooperative design optimization, and personalized user experiences. Recent developments have introduced novel frameworks and systems that enable more efficient and effective human-AI collaboration, such as multi-stage alignment frameworks for generative query suggestion and attention-aware design frameworks. These advancements have shown promising results, including improved user engagement, increased design efficiency, and enhanced user experience. Noteworthy papers include: From Clicks to Preference, which introduced a multi-stage alignment framework for generative query suggestion, and Generative Interfaces for Language Models, which proposed a paradigm for proactive generation of user interfaces. Cooperative Design Optimization through Natural Language Interaction and Insights into User Interface Innovations from a Design Thinking Workshop also presented innovative approaches to human-AI interaction. Additionally, Athena and Persode introduced novel applications of large language models to app generation and personalized visual journaling, respectively. OnGoal presented a chat interface that helps users track and visualize conversational goals in multi-turn dialogue with large language models.

Sources

From Clicks to Preference: A Multi-stage Alignment Framework for Generative Query Suggestion in Conversational System

Cooperative Design Optimization through Natural Language Interaction

Insights into User Interface Innovations from a Design Thinking Workshop at deRSE25

Generative Interfaces for Language Models

"She was useful, but a bit too optimistic": Augmenting Design with Interactive Virtual Personas

Attention is also needed for form design

Athena: Intermediate Representations for Iterative Scaffolded App Generation with an LLM

Persode: Personalized Visual Journaling with Episodic Memory-Aware AI Agent

OnGoal: Tracking and Visualizing Conversational Goals in Multi-Turn Dialogue with Large Language Models

Built with on top of