Generative AI is already woven into the industries learners are training for. Yet most vocational education and training (VET) providers and organisations are still figuring out how to respond. Blanket bans on AI use in assessments are leaving learners underprepared for AI-embedded workplaces. Trainers are making inconsistent decisions. Learners are getting mixed messages. And, the gap between what’s assessed and what’s expected in the workplace keeps widening.
The AI Assessment Scale (AIAS) offers a better approach. It helps VET providers and organisations make deliberate, consistent decisions about how AI fits into their assessments.
In this article, we explore:
- What the Artificial Intelligence Assessment Scale is and how it works
- How to apply the scale in vocational education and training (VET) and in the workplace
- Benefits, practical considerations, and implementation tips
- Common challenges and how to address them
What is the AI Assessment Scale?
The AI Assessment Scale (AIAS) is a five-level practical framework for defining and communicating acceptable AI use in assessments and tasks. It was created by Mike Perkins, Leon Furze, Jasper Roe, and Jason MacVaugh. All four are educators, professors, and consultants with a shared interest in how AI is used in student learning.
Originally developed for education, the scale helps trainers and managers set clear expectations for how much AI students and employees can use in their work. It also promotes ethical use of those tools, ensuring generative AI is critically analysed before being integrated into assessments and tasks.
For assessors and validators, the scale simplifies their role. It sets clear expectations upfront, so there is less ambiguity when reviewing work for evidence of AI use.
Why Banning AI From Assessments Doesn’t Work
Generative artificial intelligence is evolving faster than most organisations can keep up with. Many have responded by banning AI use in assessments altogether. This creates its own set of problems for learners, employees, assessors, and validators.
Banning AI from assessments raises several concerns:
- Learners miss out on developing skills they will need in their careers.
- Assessors may suspect AI use but lack reliable tools to prove it.
- Blanket bans ignore cases where AI use is entirely appropriate. Just as open-book exams test research skills rather than memory, some assessments are better served by allowing AI for specific tasks.
The AI Assessment Scale addresses these challenges directly. It provides clear guidelines on when and how to use AI in assessments. It also encourages learners to reflect on their own usage, explaining what tools they used and why. This shifts the focus from policing AI to using it purposefully.
AI Assessment Scale Levels
Note: For each of the assessment scale levels, we provide guidelines on what is and isn’t allowed. These are only suggestions, and may vary by institution. For example, some training providers could permit the use of notes, textbooks, and search engines during a No AI assessment, while others may block online access or take a paper-based essay approach.
1. No AI
No AI usage is permitted during the assessment. Learners rely solely on their existing skills and knowledge. These assessments are similar to traditional closed-book exams. They focus on measuring recall ability and skill proficiency without the aid of other references.
These assessments are particularly valuable in fields where performing without assistance is a real-world requirement.
What’s Allowed
- Personal knowledge and skills
- Formula sheets and reference cards provided during the assessment
What’s Not Allowed
- Internet access
- External resources, such as notes and textbooks
- Generative AI tools
Example
Sitting a supervised written exam, answering questions from memory alone. Another example could be demonstrating a task, such as how to safely wire an electrical circuit, under supervision.
2. AI Planning
AI may be used during the planning stage of an assessment, but not in the final submission. Learners can use AI for idea generation, outline creation, and initial research. The final work must reflect their own thinking and independent development of those ideas.
This level is valuable for teaching learners how to use AI as a starting point rather than a shortcut.
What’s Allowed
- Using AI for brainstorming, outlining, and research
- Using notes and outputs from AI planning activities during the assessment
What’s Not Allowed
- AI-generated content as part of the submission
- Lightly editing AI outputs without meaningful development
- AI assistance during the assessment process
Example
Using ChatGPT to identify relevant market trends and research sources. Then, analysing that research to build a business report reflecting learners’ own thoughts and words.
3. AI Collaboration
AI may be used to draft, refine, and evaluate assessment sections. However, learners must still critically evaluate and modify any AI-generated content they use. This ensures the final submission reflects their understanding and judgement.
This level teaches learners to recognise the limitations of AI outputs and improve on them, rather than accepting them at face value.
What’s Allowed
- Using AI to draft sections, provide feedback, and refine work
- Incorporating AI content that has been critically evaluated and meaningfully modified
What’s Not Allowed
- Submitting AI-suggested outputs without meaningful modification
- Using AI to complete the entire assessment
- Presenting AI outputs and claiming them as original work
Example
Writing a first draft of a project proposal independently, then using Google Gemini to review it for clarity, structure, and flow. Incorporating the feedback where it improves the work, and rejecting suggestions that don’t align with the students’ judgement.
4. Full AI
AI may be used extensively throughout the assessment. Learners direct AI tools to achieve their assessment goals, rather than completing tasks manually. The focus shifts from production to direction. Success is measured by how effectively a learner can prompt, guide, and evaluate AI outputs.
This level is valuable for developing in-demand AI skills needed to work in professional settings.
What’s Allowed
- Using AI to complete any elements of the assessment
- Directing AI to solve problems or achieve specific goals
What’s Not Allowed
- Submitting work without demonstrating critical direction of AI tools
- Passing off AI outputs as independent work without acknowledging AI use
Example
Directing Claude to draft, structure, and refine a full marketing strategy. Then, evaluating the output critically and adjusting it to better meet the assessment brief.
5. AI Exploration
AI is used creatively to develop innovative solutions and novel approaches to the assessment. Learners and educators may co-design the assessment itself, exploring unique AI applications within the field of study. There are no prescribed methods. The goal is to push the boundaries of what AI can do.
This level is particularly valuable for fields where AI will be used extensively and creatively in professional roles. In these industries, the ability to experiment with and direct AI tools is itself a core skill. AI Exploration also fits well into self-directed learning environments, where learners choose their own learning paths and assessment criteria.
What’s Allowed
- Co-designing assessment approaches with educators
- Using AI creatively and experimentally to see what it’s capable of achieving
- Exploring unconventional or innovative uses of AI tools
What’s Not Allowed
- Presenting AI outputs without critical reflection on the process
- Ignoring ethical considerations when experimenting with AI tools
Example
Exploring how various AI platforms can be used to generate and evaluate multiple design concepts for a product brief. Documenting the process and reflecting on where AI added value and where each AI tool fell short.
Benefits of Using the AI Assessment Scale in Education and the Workplace
Provides Clear Guidelines
The AI Assessment Scale removes ambiguity around where AI is appropriate. Each level sets out exactly what is and isn’t allowed. This gives learners, employees, and assessors a shared point of reference across higher education, VET, and workplace settings.
Promotes Responsible Use
The scale encourages a thoughtful approach to AI rather than unrestricted use. It helps learners and employees understand where AI adds value and where it introduces risk.
What Does Responsible AI Use Mean For RTOs?
Explore the legislation and approaches that Australian providers are taking to AI use across their organisations.
Supports Accountability
When combined with declaration requirements, the AIAS makes it easier to address cases where AI use has exceeded agreed boundaries. This protects institutions, training providers, organisations, and individuals.
Streamlines Validation Processes
Validators no longer need to make judgement calls about AI. The scale provides a documented framework for faster, more consistent review processes. This reduces administrative burden and strengthens audit trails.
Encourages Creative Use of Technology
The AIAS doesn’t treat AI as a threat to be managed. At higher levels, it encourages learners and employees to experiment with AI tools. This fosters innovation in university teaching, VET, and workplace learning.
Supports Career Readiness
Structured exposure to AI tools prepares learners for AI-embedded workplaces. They arrive in the workforce with practical experience and a clear understanding of appropriate AI applications. This reduces employee onboarding time and accelerates productivity.
Enables Consistent Policy Across Organisations
Without a shared, practical framework, AI policy can vary significantly between teams. The scale gives organisations a common language for managing AI. This makes it easier to implement, communicate, and enforce policy across the business.
Aligns AI Use With Business Goals
Not all AI is equally valuable. The scale helps organisations identify where AI can genuinely move the business forward. This could include analysing market data, supporting decision making, accelerating technology development, or driving adoption of new tools. This ensures AI is used purposefully, without introducing unnecessary risk.
Considerations For How to Apply the AIAS to Your Assessments
1. Desired Learning Outcomes
What do you want the assessment to achieve? If the desired learning outcome is knowledge retention or skill proficiency, Level 1 is likely the best fit. If it’s developing critical thinking, research skills, or the ability to work alongside AI, higher levels may be more appropriate.
2. Assessment Types and Methods
What type of assessment and delivery method are you using? Not every assessment method is suited to AI. Multiple choice questions leave little room for generating AI responses, making Level 1 a natural choice. Take-home essays and project-based assessments are better suited to Levels 2 or 3, where AI can support planning and refinement without replacing independent thinking.
3. Learners’ Level of Study
How far are students in their learning journey? It’s often important to establish a foundation before introducing AI into assessments. Newer learners may benefit from starting at Level 1 or 2. For advanced learners who’ve already demonstrated core competency, an appropriate level may be 3 or above. In a workplace training context, consider an employee’s existing experience and familiarity with the subject matter.
4. Marking Criteria
What are the marking criteria measuring? Assessments that look for specific correct answers are less suited to AI. Those that reward power skills like critical thinking, analysis, or creative problem solving are better candidates for higher levels.
5. Availability of AI Platforms
Do all learners have equal access to AI tools of equivalent capability? The difference between free versions of tools like ChatGPT, Claude, or Gemini may be negligible. However, access to premium versions could give some learners an unfair advantage. Factor this in before assigning a level that depends on AI access.
6. Digital and AI Literacy
Do all learners have the digital skills needed to engage with AI tools fairly? Assigning a higher AIAS level to a group with mixed literacy could disadvantage some learners and give others an unfair edge.
7. Academic Integrity
Do learners understand their responsibilities when using AI? This includes citing AI-generated content correctly, not taking outputs at face value, and verifying information independently. In a workplace context, this extends to data accuracy and avoiding over-reliance on AI where human judgement is critical.
8. Applications for AI in the Field or Industry
How widely is AI used in the learner’s field of study or target industry? In industries where AI is central to everyday work, such as software development, Level 4 or 5 may not only be appropriate, but necessary. In fields with fewer AI applications, such as nursing or trades, lower levels are likely more suitable.
Tips For Implementing an AI Assessment Scale
Ensure Compliance Alignment
Check that your use of the AIAS aligns with relevant regulatory and institutional requirements. This is especially important in VET and higher education. Involve compliance teams early to avoid issues down the line.
Get Buy-In From All Stakeholders
Introduce the scale to learners, trainers, assessors, and management before implementation. Address concerns early and communicate the benefits clearly.
Explain Where Assignments Fall on the Scale and Why
Don’t just assign a level. Explain the reasoning behind it. Learners and employees are more likely to respect boundaries when they understand the rationale.
Encourage Transparency
Create an environment where disclosing AI is the norm. When learners and employees feel they’ll be penalised for honesty, they’ll hide it. Focus on transparency as a value, not just a requirement.
Build Purposeful AI Skills to Support AI Use
Provide training on how to get the most out of AI platforms. This should take place before learners are expected to use those tools in assessments or workplace tasks.
Train Trainers, Assessors, Validators and Managers
Ensure everyone responsible for implementing and reviewing assessments understands how the scale works. Regular training maintains AI literacy and keeps understanding current as tools and policies evolve.
Offer Support Channels for Addressing AI Misunderstandings
Make it easy for learners and employees to raise concerns about AI-related decisions. This is especially important where work could be incorrectly flagged as AI-generated.
Adjust Rubrics to Factor in AI Usage
Review existing marking criteria and performance rubrics to ensure they reflect the level of AI use permitted. For example, a rubric designed for a No AI assessment isn’t appropriate for the AI Collaboration level or above.
Challenges and Limitations of the AI Assessment Scale
The Scale Can Feel Rigid in Practice
Some assessments naturally span more than one level on the assessment scale. Forcing them into a single category can feel reductive.
Solution: Allow hybrid designations where appropriate. For example, include a note that an assessment could be considered level 2 or 3 with appropriate declarations and justifications. Provide assessors with clear guidance on how to handle boundary cases consistently.
AI Capabilities Change Faster Than Policy
What counts as AI assistance today may look very different in six months. Policies built around specific tools or capabilities can quickly become outdated.
Solution: Revisit scale definitions and assessment levels at least once per year or review period.
Reliance on Transparency and Accountability
AI detection tools can be unreliable. A framework that depends on catching dishonesty is harder to sustain than one that encourages honesty.
Solution: Include declaration fields where learners document which tools they used and how. Build a culture where honest disclosure is expected and rewarded.
How Cloud Assess Supports the AI Assessment Scale (AIAS)
Cloud Assess is a training and assessment tool designed for training providers and skill development. Its features make it easier to leverage AI in training and implement the AIAS in practice.
The platform’s AI Marking Assistant ensures assessments are reviewed consistently across all levels of the scale. This reduces the risk of bias or inconsistency between assessors. Built-in plagiarism detection also supports academic integrity, flagging submissions that may not reflect a learner’s own work. This can be tracked against declarations and the assessment’s AIAS level.
Cloud Assess’ Skills Matrix also helps organisations track AI literacy across their workforce. It identifies gaps and informs targeted training decisions.
Together, these features make it easier to manage AI use purposefully, at every level of the scale. But don’t take our word for it! Schedule a demo to see Cloud Assess features in action.
AI & Automation
