About This Framework
Understanding AI fluency as the ability to work effectively, ethically, and safely with emerging AI tools.
AI Fluency is defined as the ability to work effectively, efficiently, ethically, and safely within emerging modalities of Human-AI interaction.
The Framework for AI Fluency describes the interconnected competencies needed to use AI in creative, innovative, and problem-solving work. Rather than viewing AI merely as an efficiency engine, the framework recognizes AI's potential as an authentic thinking partner for meaningful cognitive work — while acknowledging that this potential is only realized through the development of specific human competencies.
This framework emerged from research collaboration across higher education and extends to organizations addressing the challenges and opportunities of generative AI. DC CAP's adoption of this framework reflects its alignment with our enterprise AI adoption principles and our commitment to building genuine organizational AI fluency.
Key Advantages
Platform Agnostic
Independent of specific tools or platforms. Adaptable to rapidly evolving technologies and use cases.
Contextual & Flexible
Characterizes effective action rather than prescribing rigid processes. Compatible with other skills taxonomies.
Ethics-Centered
Treats ethical considerations as fundamental. Responsible and safe AI use is as important as responsible AI design.
Three Interaction Modalities
Human-AI interactions often bridge multiple modalities, with practitioners moving between contexts even within single projects.
AI performs tasks independently, based on direct human instructions. Particularly useful for improving the efficiency of repetitive, time-consuming, or data-intensive tasks.
- Requires clear task definition and quality control measures
- Focuses on efficiency gains and task completion
AI and human co-define and co-execute tasks in an iterative way, collaborating toward an end goal. Focuses on enhancing human expertise rather than replacing it.
- Dynamic interplay between human and AI contribution
- Involves dialogue and iterative refinement
Human configures AI to independently perform future tasks (including for others) on behalf of the user. Defines the characteristics and behavior of an AI.
- Requires sophisticated understanding of AI capabilities and limitations
- Enables at-scale deployment of AI assistance
Core Competencies: The 4Ds
Four interconnected competencies that enable effective, efficient, ethical, and safe Human-AI interaction.
The 4Ds reinforce each other. Delegation requires Discernment to evaluate tool choices. Description requires Delegation to know what to ask for. Diligence applies to all three.
Delegation
Creative vision and selection of the right AI tools.
The ability to identify when and how to use AI tools and modalities effectively in creative and problem-solving processes. Understanding capabilities and limitations of AI technologies and making informed decisions about when to use Automation, Augmentation, or Agency.
- Envisioning an effective goal for a project
- Understanding the nature and requirements of tasks toward the defined goal
- Analyzing and deconstructing a task into AI, human, and collaborative components
- Understanding the capabilities and limitations of current AI tools
- Knowledge of various AI platforms and their specific strengths
- Evaluating tools based on project requirements, budget, and operational needs
- Balancing AI and human capabilities throughout a project
- Understanding different affordances of each modality
- Assigning project tasks to human and AI tools optimally
Description
Effectively describing tasks to prompt AI behaviors.
The skills needed to effectively communicate ideas, requirements, constraints, and other aspects of creative visions to AI systems. Crafting clear, specific, and well-structured prompts that guide AI to produce desired behaviors and outputs.
- Prompting to define desired output
- Clearly articulating desired characteristics and features
- Translating creative vision into AI-understandable terms
- Dialogic prompting to produce effective iterative collaboration
- Engaging in dynamic, back-and-forth communication with AI
- Breaking down complex tasks into manageable prompts
- Directive prompting to define future AI behaviors
- Defining how AI-generated content should behave in the world
- Translating user needs into AI behavior guidelines
Discernment
Accurately assessing the usefulness of AI outputs.
The critical evaluation of AI-generated outputs — understanding their quality, relevance, potential biases, and other characteristics. The ability to iterate and refine the collaborative process with AI tools.
- Evaluating output quality and identifying improvements
- Critically assessing relevance and effectiveness
- Identifying strengths and weaknesses in AI outputs
- Assessing if human-AI collaboration is fruitful
- Evaluating effectiveness of the collaborative dynamic
- Identifying which aspects are most beneficial and where improvements fit
- Evaluating if AI-driven independent behaviors enable positive experiences
- Assessing effectiveness in user-facing scenarios
- Gathering and interpreting feedback to refine AI behaviors
Diligence
Taking responsibility and vouching for final products.
The responsible use of AI — including ethical considerations, transparency about AI use, and taking accountability for final products created with AI assistance.
- Responsible use of AI tools with ethical and legal best practices
- Understanding and applying ethical principles throughout the process
- Identifying and mitigating potential biases and ethical risks
- Transparency and accountability when distributing end products
- Understanding audience, industry, and legal expectations
- Clearly communicating the nature of AI involvement
- Taking responsibility for verifying and vouching for AI outputs
- Implementing thorough fact-checking and testing procedures
- Managing and assuming responsibility for deployment risks
DC CAP Application Notes
How the 4Ds map to DC CAP's enterprise AI adoption strategy.
The 4Ds map directly to DC CAP's enterprise AI adoption strategy and Anthropic's framework for responsible AI use. Every staff member with a Claude Enterprise seat is expected to develop competency across all four dimensions.
| 4D Competency | DC CAP Application |
|---|---|
| Delegation | Choosing when to use Automation (email drafts), Augmentation (grant writing), or Agency (custom Skills) for each workflow |
| Description | Writing effective prompts, Project Instructions, and Skill configurations that produce on-brand, policy-accurate outputs |
| Discernment | Reviewing Claude outputs for accuracy, appropriate tone, correct data framing, and policy compliance before any external use |
| Diligence | Following the AI Governance Framework's data classification tiers, attribution norms, and incident response protocols |
Phased Adoption Path
The three interaction modalities also map to the phased adoption plan: Phase 1 focuses primarily on Automation and Augmentation. Phase 3 introduces full Agency through Skills, Connectors, and multi-agent workflows.