Researchers at UC San Diego introduced Orca, an open-source system that enhances web interactions by guiding users with large language models. Unlike autonomous agents, Orca serves as a 'co-pilot,' improving task speed and accuracy. The system offers features like summarizing web content, extracting structured data, and reducing context-switching. A study with participants showed enhanced web exploration and trust in the AI's assistance. Orca's design promotes shared control, ensuring users remain in command, which is vital for transparency and confidence in using AI solutions for real-world workflows.
Orca is designed to assist users in extracting insights from the web without taking control, exemplifying a decision-making co-pilot for improved task performance.
The system allows users to delegate context-rich tasks while staying in control, emphasizing shared control as key for trust and confidence in workflows.
Collection
[
|
...
]