We invite submissions (full papers and extended abstracts) for the first Symposium on AI Risks in Education and Safeguards (AIRES 2027).

AIRES provides an interdisciplinary forum for researchers, educators, policymakers, students, quality assurance agencies, civil society organizations, and industry representatives to explore the opportunities and risks of artificial intelligence in education. The symposium focuses on safeguarding educational values while addressing challenges related to integrity, human agency, fairness, accountability, data governance, and digital sovereignty.

Areas of interest

We encourage submissions from researchers and practitioners across education, learning sciences, computer science, philosophy, social sciences, law, policy, and related fields. Interdisciplinary and transdisciplinary contributions are particularly welcome, especially those that integrate multiple perspectives or are developed in collaboration with relevant non-academic stakeholders.

A non-exhaustive list of themes includes:

Theme 1: Trust, Assessment, and Academic Integrity

How do we know learning has actually occurred in AI-enabled educational environments?

Generative AI is challenging traditional approaches to assessment, authorship, and competency verification. This theme explores how institutions can maintain trust in learning outcomes and qualifications while embracing responsible uses of AI.

Topics include:

• Academic integrity in the age of AI

• AI-assisted learning versus Academic cheating with generative AI

• Assessment redesign and authentic assessment

• Competency verification

• Oral examinations and competency assessment

• AI-supported assessment and testing

• Trust in qualifications and credentials

• Educational standards and quality assurance

• Academic misconduct, plagiarism, and misinformation

• Human oversight in assessment processes

Theme 2: Human Agency, Curiosity, and AI Literacy

What happens when AI becomes a cognitive partner in learning and teaching?

As AI increasingly supports thinking, writing, problem-solving, and decision-making, important questions emerge regarding human judgement, curiosity, dependency, and the future of learning.

Topics include:

• The Curiosity Paradox

• Cognitive offloading and cognitive debt

• Dependency risks

• Critical AI literacy

• Student agency and learner autonomy

• Teacher agency and professional judgement

• Human-AI collaboration

• Human oversight

• Metacognition and self-regulated learning

• Future educator roles

• Future graduate capabilities

• AI acceptance and adoption in education

Theme 3: Fairness, Inclusion, and Educational Justice

Who benefits from AI in education, and who may be disadvantaged?

AI systems can support inclusion and personalization but may also reproduce existing inequalities and biases. This theme examines how educational institutions can promote fairness and educational justice in AI-enabled environments.

Topics include:

• Algorithmic bias and discrimination

• Educational equity and inclusion

• Accessibility and universal design

• Language and cultural inequities

• Neurodiversity and AI-supported learning

• Representation in AI systems

• Profiling and discrimination risks

• Fair use of student data

• Ethical implications of educational AI

• Inclusive design and responsible innovation

Theme 4: Data Governance, Learning Analytics, and Digital Sovereignty in Education

How can educational institutions govern data, analytics, and AI infrastructures responsibly while protecting privacy, sovereignty, and educational values?

Educational institutions increasingly rely on data-intensive systems, learning analytics, educational data mining, and AI-enabled platforms to support teaching, learning, assessment, and decision-making. These developments raise important questions about governance, privacy, transparency, accountability, digital sovereignty, and the role of data-driven systems in education.

This theme examines how educational institutions can responsibly govern educational data, analytics, and AI-enabled infrastructures while protecting learners, safeguarding public values, and maintaining institutional autonomy.

Topics include:

• Learning analytics

• Educational data mining

• Student data governance

• Privacy and data protection

• Adaptive and personalised learning

• Student profiling

• Algorithmic decision-making

• Educational data infrastructures

• Digital sovereignty

• Educational sovereignty

• Vendor lock-in and platform dependency

• Open educational infrastructures

• Data sharing and interoperability

• Explainability and transparency

Theme 5: Governance, Accountability, and Responsible AI

How can educational institutions deploy and govern AI responsibly?

The successful adoption of AI requires effective governance, accountability mechanisms, quality assurance frameworks, and institutional capacity.

Topics include:

• Institutional AI governance

• Responsible AI deployment and monitoring

• AI policies and governance frameworks

• Procurement and vendor accountability

• Transparency and explainability

• Accountability and oversight

• Quality assurance and accreditation

• Certification and quality marks for educational AI

• Institutional readiness

• Teacher and institutional capacity building

• Regulatory frameworks

• Evidence-based AI adoption

• Responsible EdTech innovation

Interdisciplinary and cross-cutting submissions spanning multiple themes are particularly encouraged. 

Evaluation criteria

Papers will be judged based on scholarly criteria in their subject area, including

• Relevance to the symposium themes (see areas of interest above).

• Quality and clarity: correctness, comprehensiveness, depth of exposition, methodological soundness, evaluation of both strengths and limitations, etc.

• Originality: the work presents new analyses, tasks, methods, etc., new perspectives on existing work, or combines existing work in a novel way.

• Potential for impact: the potential to influence academic discipline, public discourse, policy, or real-world systems.

AIRES prioritizes high-quality contributions that advance understanding of AI risks and safeguards while supporting responsible, equitable, and human-centered approaches to AI in education. We will have reviewers from different disciplines who bring the expertise to evaluate work from all the areas we solicit from.

Submission format

Authors can choose between submitting a full paper or an extended abstract. At least one author of each accepted submission is expected to be present at AIRES 2027 to present their work. Additionally, we expect authors to be available to review other submissions in their area (light load).

Full paper (long)

This option is suitable for contributions covering novel work. You will receive formal peer reviews for your paper and, if accepted, a chance to present your work at the symposium as well as a (citable) full paper in the symposium proceedings.

• Length: A full paper of up to 15 pages (plus unlimited pages for references and appendix). Papers that do not follow the length requirement may be rejected without review.

• Dual submission policy: It is not permitted to submit work that is already under review, has been accepted for publication, or has already been published at another venue as a full paper.

• Anonymization: Full papers are reviewed mutually anonymously (i.e. reviewers do not know the names of the authors and authors do not know the names of reviewers). Authors must omit names, affiliations, and other potentially identifying information from the submission. Citations to prior work from the authors should be made in the third person. Non-anonymized papers may be rejected without review.

• Supplementary materials. Supplementary materials can be included as an appendix in the submitted manuscript. Reviewers are not obliged to look at it or consider it in their assessment of the submission.

Extended abstract (short)

This type of submission is flexible and suitable for a range of submissions, including but not limited to ongoing/early-stage work, work under review elsewhere, already published work, case studies from industry, and open challenges in research or practice. You will receive succinct peer reviews for your paper and, if accepted, a chance to present your work at the symposium.

• Length: An extended abstract of up to 4 pages (plus unlimited pages for references).

• Dual submission policy: It is permissible to submit work that is under review, has been accepted for publication at another venue, or has been published at another venue (provided this is in line with the submission policy of the other venue).

• Anonymization: the author's name and affiliation shall appear in the submission. The review is single blind..

Archival and non-archival

Authors can choose between archival and non-archival submissions:

• Archival: accepted papers (both full papers and extended abstracts) shall be submitted to Proceedings of Machine Learning Research (PMLR) for publication in the online symposium proceedings.

• Non-archival: authors can opt-out from publication in PMLR proceedings.

Formatting instructions

The symposium proceedings will be published online as a volume in Proceedings of Machine Learning Research (PMLR). Papers must be submitted as a PDF file on OpenReview, formatted according to the AIRES template. Templates are available below for Word, LaTeX, and Overleaf users.

• Word template 

• LaTeX template

• Overleaf template 

Presentation of Accepted Work

Regular presentation

The authors of each accepted paper can choose to present their work in a lightning round and/or poster session. Regular presentations are aimed at getting a sense of the work happening within our community and encouraging participants to make new connections.

• Lightning round: short presentations (5 minutes) in which one of the authors summarizes the work.

• Poster session: posters summarizing the work are displayed in the poster presentation area throughout the symposium. Each poster is presented by one of the authors during one of the poster sessions.

In-depth presentation

A set of papers will be selected for an in-depth presentation (20 minutes + room for questions). Both short and long submissions will be considered for the in-depth track based on fulfilling the overall goals of the symposium. The quality of the paper is certainly a factor, but selection for the in-depth track results from a plurality of considerations, including achieving a balance between the representation of diverse views and disciplinary perspectives and fostering interdisciplinary dialogue.

Contact

For questions, please contact general-chairs@aires-symposium.org