The Future of Assessment: A Blueprint for Learner Evaluation in an AI World

Introduction

The recent widespread availability of generative AI has thrown into question all assessment processes, including tasks like essays and exams which have long formed the basis of learner evaluation. As institutions attempt to outline an updated approach to assessment that responds to the disruption of AI and the constantly changing modern economy, this paper is designed as a helpful starting point: we’ve endeavoured to produce a grounded summary of current assessment trends and how institutions can best position themselves to proactively respond to change. This paper is broken down into two main sections. In the first section, we look at reasonable assumptions that institutions can make regarding the future of education.

In the second section, we look at actionable steps that institutions can take today to prepare for an AI-augmented educational and professional environment.

Reasonable assumptions for the future

Let’s be clear: there is (sadly) no crystal ball here at Anthology HQ that can predict the future with absolute accuracy. It is also clear, however, that uncertainty shouldn’t form an excuse for inaction. To lay the groundwork, here are key assumptions we believe can be reasonably made—all of which have a fundamental impact on how higher education should approach assessment.

1. AI is quickly becoming ubiquitous

AI is going to significantly impact all major industries, with the list of the least affected jobs already reduced to those in the purely physical realm, such as the roles performed by professional athletes, dancers, and roofers. As Ethan Mollick notes, nearly every occupation now overlaps with AI in some way.

Assessment has followed a similar trajectory. Studies across a range of academic disciplines have shown that newer large language models (LLMs) are able to complete common academic tasks such as exams to a very high standard. These capabilities are already being used daily by millions of learners, a figure that will continue to grow rapidly. As a result, we must be preparing for a world

where learners have access to tools that can approximate human intelligence across all areas of reasonable academic inquiry.

2. AI must have an active role in pedagogy

What is required now is a move towards co-intelligence, wherein our role as educators is to fuse AI with existing pedagogical techniques to develop the knowledge and skills that allow students to succeed in the modern world. In recent years, a vast amount of research has been dedicated to understanding whether using AI can improve student performance, with results indicating that it can.

Leading institutions are also observing this trend in their courses. As Stephanie Richter, Director of Teaching Excellence and Support at Northern Illinois University (NIU) and member of Anthology’s Future of Assessment Working Group, explains, the use of AI tools in Blackboard is helping NIU’s faculty to create inspiring and engaging courses. “The results from the AI Design Assistant have given me new ideas for assessments that I had not previously considered,” Richter notes.

3. Assessment will involve a program, not just a task

In a world where AI is ubiquitous, it won’t be possible to have a single, holistic task that accurately evaluates student understanding and proficiency. Assessment is much better thought of as a program that allows the instructor to gauge the learner’s competence over time, rather than a stand-alone activity. AI’s emergence will require institutions to adopt formative assessment more broadly and employ a broader range of activities in order to truly understand student progress.

4. A changing workforce will require a more flexible assessment approach

Technological change means that employable skills are being recycled with increased frequency. If we’re to conceive of higher education assessment as a means of evaluating a student’s readiness to enter the workforce, the implication is clear: assessment will need to be more regularly updated to account for these changes in industry. Developing the flexibility to respond to market needs will be a fundamental component of the future of assessment.

Five steps to take today to prepare for the future of assessment

With the above context in mind, here are some key initiatives institutions should investigate in the near term to evolve their approach to assessment and meet the needs of the AI era.

1. Immediately address AI literacy among instructors

Currently, large numbers of faculty are less confident using AI than the students they’re trying to teach and are thus resistant to evolving their courses for the AI world. No institution can consider their assessment practices future-proofed while this remains the case.

Late in 2024, we embarked on the Ethical AI in Action World Tour, where we visited 25 global cities and spoke with education leaders about the opportunities and risks AI presents. The instructors we spoke with expressed concerns around a lack of clarity on AI policies at their institutions, and a hesitancy to engage with generative AI in a pedagogical context.

2. Embrace—and expand—authentic assessment practices

We believe authentic assessment can be more than a strategy to avoid AI plagiarism while also forming a crucial component of an effective assessment program in the AI world. So how can authentic assessment be fused with the benefits of AI instruction to create a revised approach to pedagogy?

Lisa A. Clark, EdD, associate vice president for academic transformation here at Anthology, has outlined our response in her recent whitepaper, Reframing Bloom’s for the Age of AI: A Whitepaper for Future-Ready Educators. Clark’s work revisits the pedagogical basis of authentic assessment, Bloom’s Taxonomy, and outlines how it can be evolved to suit a world in which generative AI is ubiquitous. It reviews each established level of the taxonomy and provides a suggested reframing, before arguing the need for a seventh layer, transform, in which the student is required to leverage human-AI collaboration to drive meaningful real-world impact.

3. Align skills attainment with the workforce pipeline

With industry increasingly prioritising skills attained above degrees held, higher education must grow to meet these needs and set learners up for success. This involves two crucial components: firstly, adopting assessment practices which allow learner skills to be evaluated, recognised, and shared; and secondly ensuring that the skills being taught align with market demand.

In a skills-driven economy, the way capabilities are signaled is as important as the skills themselves. Micro-credentials offer one of the most agile responses, providing recognition for specific competencies that can be earned quickly and stacked toward larger qualifications (Bozkurt et al., 2023). Work-integrated learning—from internships to live-client projects—gives learners the opportunity to apply knowledge under conditions that mirror professional practice.

Here again there is a significant role for instructor training and development, as faculty can only create courses that meet market needs if they’re provided the latest data on what those needs are.

We recently conducted the 2025 Anthology Faculty Survey, where responses from thousands of instructors revealed many key opportunities for institutions to provide better support and enhance teaching accordingly. Fueling the workforce pipeline is one of the key areas: instructors consider “meeting the demands of a rapidly changing job market” as the second biggest challenge facing higher education, and yet only 19% report being “very confident” that their courses align with the latest workforce trends.

4. Consider opportunities for adaptive learning to personalise instruction

In addition to skills-based learning, institutions should also consider the possibility of an adaptive delivery model. Adaptive learning involves leveraging AI to first evaluate a student’s current level of proficiency, and then to tailor their learning path based on the areas where improvement is required. This allows institutions to do personalised learning at scale, and learners to efficiently build the skills they need to succeed.

Over the last 12 months, we partnered with Obrizum to do a pilot study of adaptive learning in Blackboard®. We found that the key benefit came through time savings. The most advanced students were able to complete their course up to 30% faster, as they weren’t required to spend long periods on areas they already understood. For students nearer the mean, the saving was closer to 15%.

The pilot also revealed that adaptive learning requires a change of culture, not just the addition of new technology. This includes a larger content pool to allow AI the flexibility to tailor a path for each learner; increased emphasis on MCQs; and further training for instructors.

5. Leverage data to inform assessment design

Data sits at the core of a future-proofed assessment approach. The rise of generative AI will supercharge this space, including more scope to:

1) Understand how different students interact with learning content, and automate personalised learning paths

2) Analyse assessment and broader course design tasks in depth, covering not just student performance but a more in-depth view of task design and opportunities to optimise

3) Leverage AI to develop a broader pool of assessment tasks

4) Combine academic data with other data sources to provide a holistic view of student progress

5) Easily mine this data to unlock insights, with natural language queries and actionable responses that empower instructors to take action

Future of Assessment – A Summary

Reasonable assumptions for the future:

1) AI will become ubiquitous

2) AI must have an active role in pedagogy

3) Assessment will involve a program, not a task

4) A changing workforce will require a more flexible approach

Five steps to take today to prepare for the future of assessment:

1) Immediately address AI literacy among instructors

2) Embrace and expand authentic assessment practices

3) Align skills attainment with the workforce pipeline

4) Consider opportunities for adaptive learning to personalise instruction

5) Leverage data to inform assessment design

Written for OEB 2025 by Blackboard by Anthology