Explain the 10/80/10 paradigm when applying generative-AI

The 10/80/10 paradigm is a framework used in applying generative-AI to tasks or projects, emphasizing the division of effort and responsibility between AI and human experts. This paradigm typically works as follows:

1. The First 10%: Defining the Problem and Initial Inputs

• Human Responsibility: Humans clearly define the problem, set goals, and provide the AI with necessary context and constraints.

• Focus: This stage involves crafting high-quality prompts, ensuring that AI has the relevant data, and clarifying expected outputs.

• Importance: This phase is crucial for setting up AI for success by providing clear guidance on what is required.

2. The 80%: Generative-AI Execution

• AI Responsibility: The generative-AI performs the bulk of the work, creating drafts, analyzing data, generating content, or designing solutions.

• Automation: AI applies its models to generate initial outputs quickly, based on the inputs from the first phase.

• Iterative Feedback: AI may refine its outputs through iterative prompts or feedback provided by humans during this stage.

• Scope: This step maximizes the efficiency and capabilities of AI to do the heavy lifting, providing a comprehensive foundation.

3. The Final 10%: Refinement and Validation

• Human Responsibility: Humans critically evaluate, refine, and finalize the AI-generated outputs to ensure quality, accuracy, and alignment with the original goals.

• Verification: This includes fact-checking, aligning the results with nuanced requirements, and ensuring compliance with standards or ethics.

• Polish: Humans add creativity, strategic insight, or domain expertise that AI might lack, making the final product robust and effective.

Key Benefits of the 10/80/10 Paradigm

• Efficiency: Leverages AI's strengths to handle repetitive, data-intensive tasks while reserving human effort for critical strategic thinking and validation.

• Collaboration: Balances AI and human contributions, combining automation with human expertise for superior outcomes.

• Scalability: The approach can be scaled across industries and tasks, from creative work to technical problem-solving.

Example Application

• Writing a Report:

  1. 10% Initial Input: A human outlines the purpose, target audience, and key points to cover.

  2. 80% AI Generation: AI drafts the report, integrating data, writing summaries, and structuring the content.

  3. 10% Final Refinement: A human edits the draft for tone, accuracy, and readability, adding unique insights and ensuring alignment with organizational goals.

This paradigm ensures that AI and humans work in synergy, optimizing the strengths of both.