Skip to content
Live
Trending storyPast story

Google Research released the Agile Deliberation framework

2 reports2 outlets14 days agoUpdate

The story so far

AI overview

Google Research has launched the Agile Deliberation framework, which assists humans in defining the boundaries of subjective concepts through iterative dialogue, transforming vague ideas into precise AI instruction blueprints. The framework is divided into two stages: concept definition and concept iteration. In 18 user experiments of 1.5 hours each, it significantly reduced user fatigue and anxiety, making the performance of AI models better than standard baselines. Google Research has open-sourced the code and algorithms and conducted 18 user experiments in random order to prevent learning bias. The framework is based on the Gemini 2.5 Pro and Gemini 2.5 Flash API, requiring no fine-tuning, with an average F1 score 10.5 points higher than zero-shot Gemini 2.5 Flash, reaching 0.59 in paid scenarios.

AI-written from the reports · Updated 14 days ago

Coverage timeline

Follow the coverage to see every side of the story.

5 Nov
  1. MarkTechPost
    Google AI Research Releases Agile Deliberation Framework for Gray-Area Image Classifiers

    Google AI Research releases Agile Deliberation framework, which transforms fuzzy concepts into structured definitions through human-in-the-loop, for gray-area image classification. The framework is based on Gemini 2.5 Pro and Gemini 2.5 Flash API, requiring no fine-tuning, with an average F1 score 10.5 points higher than zero-shot Gemini 2.5 Flash, reaching 0.59 in paid scenarios.

  2. Google Research
    Agile Deliberation: How AI helps humans define the gray areas

    The Agile Deliberation framework assists humans in defining the boundaries of subjective concepts through iterative dialogue, transforming fuzzy ideas into precise AI instruction blueprints. The framework is divided into two stages: concept definition and concept iteration, which processes fuzzy images through direct feedback. In 18 user experiments of 1.5 hours each, it significantly reduces user fatigue and anxiety, making AI model performance superior to standard baselines.