NASA’s new dark energy space telescope can also detect killer asteroids
technologyreview.com·2h ago
AI Research
arXiv CS.AI·Jun 12·5 min read
Interaction-Centered Intelligence: Toward an Interaction-Based Theory of Human-AI Co-Creation
TL;DR
Traditional AI often overlooks the importance of interaction in intelligence development. This paper introduces Interaction-Centered Intelligence, a framework that emphasizes the role of interaction in human-AI co-creation.
✦ Why It Matters
Engineers can leverage interaction dynamics to enhance human-AI collaboration and improve system adaptability.
Key Takeaways
- Focus on interaction rather than just outputs in AI systems.
- Explore frameworks that enhance human-AI co-creation.
- Utilize interaction trajectories to assess AI effectiveness.
- Consider implications for explainable and hybrid intelligence.
Full Summary
Traditional artificial intelligence has focused on isolated computation within individual models, often neglecting how interaction contributes to intelligence and creativity. This paper proposes Interaction-Centered Intelligence as a new framework that prioritizes interaction as the primary unit of analysis for co-creative AI.
Drawing from concepts like distributed cognition and participatory sense-making, it highlights how intelligence emerges from the dynamics between agents, environments, and socio-technical systems. The research builds on previous work in Creative Sense-Making and co-creative systems, such as the Drawing Apprentice and AI Drawing Partner.
By emphasizing interaction trajectories and coordination patterns, the framework offers a more nuanced understanding of human-AI collaboration. Implications for explainable AI and hybrid intelligence systems are also discussed, suggesting a shift in how we evaluate AI's effectiveness.