Critical Evaluation is one of the five components of AI literacy in the ALiF Standard. Every conforming implementation addresses all five, with emphasis that varies by professional role.
Key elements
- Output verification methods
- Quality assessment approaches
- Bias identification
- Impact measurement
Example competencies
- Verifies accuracy of AI-generated content
- Evaluates effectiveness of AI tools for specific purposes
- Identifies potential biases in AI outputs
- Measures impact of AI use in professional contexts
How it relates to the other components
Technical Understanding (TEC) and Critical Evaluation (EVA) as reciprocal. Understanding how AI tools work (TEC) informs the ability to evaluate their outputs (EVA), and the process of evaluating outputs deepens understanding of how tools function. These components develop most effectively when paired rather than sequenced.
Sub-competencies (3)
EVA.1 Output Assessment
The ability to verify the accuracy, completeness, and reliability of AI-generated content. This sub-competency encompasses fact-checking AI outputs against authoritative sources, identifying hallucinations and fabricated information, detecting errors in AI-generated reasoning, and validating the credibility of AI-produced information. It is the first line of defense against uncritical acceptance of AI outputs.
Proposed in ALiF 2.0: Adds checking the authenticity and provenance of AI-generated or AI-altered media (text, images, audio and video) met at work, whether the professional produced it with AI or received it from others.
| Foundation (L1) | Intermediate (L2) | Advanced (L3) |
|---|---|---|
| Verifies AI-generated content against reliable sources. Identifies obvious errors and hallucinations. Uses basic fact-checking strategies. Recognizes when output requires further verification. | Implements systematic verification procedures. Develops efficient strategies for different output types. Identifies subtle errors and inconsistencies. Creates verification checklists for routine tasks. | Establishes verification standards for teams or organizations. Designs verification protocols that balance rigor with efficiency. Trains others in assessment methodology. Evaluates and improves existing verification systems. |
EVA.2 Quality Control
The ability to systematically assess the overall quality of AI-generated work and the performance characteristics of AI tools. This sub-competency encompasses evaluating content quality beyond factual accuracy (coherence, completeness, appropriateness), assessing how well an AI tool performs for a given purpose, and recognizing biases and blind spots in AI system behavior. Where EVA.1 asks “is this output correct?”, EVA.2 asks “is this output good enough for its intended purpose?”
Proposed in ALiF 2.0: Adds recognizing tasks that a tool does not handle reliably, and how AI ranking, filtering and recommendation shape which information reaches the professional. Clarifies the boundaries with TEC.1 and EVA.1.
| Foundation (L1) | Intermediate (L2) | Advanced (L3) |
|---|---|---|
| Assesses basic quality of AI-generated content. Uses simple criteria to judge output appropriateness. Identifies common biases in AI responses. Recognizes when output quality is insufficient. | Implements structured quality assessment procedures. Evaluates AI tool performance against defined benchmarks. Develops bias detection strategies. Creates quality metrics for AI-assisted work. | Establishes quality standards that account for context-specific requirements. Designs assessment frameworks that integrate multiple quality dimensions. Leads quality improvement initiatives based on systematic performance data. |
EVA.3 Impact Analysis
The ability to measure and evaluate how AI use affects professional outcomes, efficiency, and quality. This sub-competency encompasses tracking the effects of AI integration on task completion, work quality, and productivity; identifying both benefits and unintended consequences; and using evidence to inform decisions about continued or modified AI use. It connects evaluation to organizational learning.
Proposed in ALiF 2.0: Adds the effects of AI use on the opportunities through which professionals, and newcomers in particular, develop and maintain their expertise.
| Foundation (L1) | Intermediate (L2) | Advanced (L3) |
|---|---|---|
| Tracks basic effects of AI use on task completion and quality. Identifies obvious benefits and drawbacks. Documents AI’s influence on work outcomes. | Analyzes efficiency gains and quality changes from AI integration. Develops measurement approaches for AI impact. Compares outcomes with and without AI assistance. Uses impact data to inform workflow decisions. | Designs multi-variable impact studies across complex workflows. Creates organizational measurement systems for AI value assessment. Connects impact evidence to strategic decisions about AI investment and adoption. |
Definitions and level descriptors are quoted from ALiF Standard v1.0r1. The level descriptors are informative; role-specific descriptors are in the Competency Framework.