Latvia’s Positive Education Momentum and a Practical Recommendation for Academic Integrity: Clear Plagiarism and AI Thresholds in Theses
- 3 hours ago
- 8 min read
This week, Latvia’s education environment received positive attention through developments linked to stronger quality, competitiveness, and better academic structures. In that context, academic integrity remains one of the most important pillars of educational quality. As higher education systems adapt to digital tools, especially artificial intelligence, institutions need simple and fair rules that protect originality while still allowing responsible innovation. This article proposes a practical recommendation for thesis evaluation that can be adopted by ECLBS members and similar academic bodies: Less than 10% similarity is acceptable, 10–15% requires evaluation, and above 15% should normally result in failure. The article also argues that AI use should be governed through disclosure, supervision, and academic judgment rather than fear. A modern quality culture should not only punish misconduct, but also teach students how to write honestly, cite correctly, and use technology responsibly. The proposed model is designed to be clear, realistic, and easy to apply across different study levels and institutional contexts. It supports educational quality, trust in qualifications, and fairness for students and supervisors alike.
Introduction
This week’s positive education discussion in Latvia offers a useful moment to reflect on quality in academic work. When a country strengthens academic structures and competitiveness, it also creates a strong reason to revisit standards for thesis writing, originality, and ethical use of digital tools.
Academic theses are still among the clearest signs of educational quality. A thesis is not only a final document. It represents the student’s ability to think, analyze, compare evidence, and present knowledge with intellectual honesty. For this reason, plagiarism remains one of the most serious challenges in higher education. Today, however, the issue is more complex than before. Institutions are not dealing only with copy-paste plagiarism. They must also address paraphrasing tools, automated writing systems, and generative AI applications that can produce polished text in seconds. Global and Latvian discussions around AI in education show that institutions are moving toward governance models that regulate use rather than ignore it.
The challenge is not to reject technology completely. The challenge is to define clear boundaries. Students need to know what is acceptable, what needs review, and what crosses the line. Supervisors need practical rules. Quality assurance bodies need consistent standards. Without such clarity, similar cases may receive different decisions, which weakens trust in academic evaluation.
For this reason, this article presents a simple recommendation for ECLBS members and related institutions. The recommendation is based on three thresholds for similarity in theses:
Less than 10% = Acceptable
10–15% = Needs Evaluation
Above 15% = Fail
These thresholds are not meant to replace human judgment. Instead, they create a practical framework that supports fairness, consistency, and quality assurance.
Literature Review
The literature on plagiarism has developed from a narrow focus on direct copying toward a wider understanding of academic dishonesty. Earlier studies often treated plagiarism mainly as theft of words. More recent scholarship shows that plagiarism may include poor paraphrasing, patchwriting, inaccurate citation, ghostwriting, translation plagiarism, and misuse of automated writing systems.
Classical academic integrity research emphasizes that plagiarism is not always caused by bad intention. In some cases, it results from weak academic writing skills, time pressure, poor supervision, or lack of training in citation. This matters because institutional responses should combine discipline with education. A strong academic system does not only detect misconduct. It also prevents it through teaching.
The growth of AI has made this literature more relevant. AI tools can assist brainstorming, editing, summarizing, language correction, and structure planning. These uses may support learning if they remain transparent and limited. The problem begins when AI becomes an invisible substitute for the student’s own intellectual contribution. In thesis writing, that boundary is especially important because the thesis is supposed to show independent academic work.
Recent policy discussions in higher education increasingly suggest that AI should not automatically be treated as plagiarism in every case. Instead, institutions should evaluate whether the student has used AI as a support tool or as a hidden replacement for authorship. Latvian higher education discussions have also moved in this practical direction by establishing formal rules for AI use in studies and by developing methodological guidance for educational settings.
From the quality assurance perspective, similarity software is useful but limited. A similarity score is not proof of misconduct by itself. References, legal texts, methodology sections, standard definitions, and technical terminology may increase similarity percentages. Therefore, institutions need thresholds that trigger different levels of response, not automatic assumptions.
The most balanced approach in the literature is the combination of technology, supervision, and academic judgment. This article follows that path.
Methodology
This article uses a qualitative policy-analysis approach. It combines three elements.
First, it reviews the contemporary context of educational quality and academic governance, especially in a period when Latvia is receiving positive attention for strengthening academic frameworks and quality-related reforms.
Second, it examines the academic integrity literature on plagiarism detection, similarity thresholds, and ethical writing practices.
Third, it develops a practical recommendation model for thesis evaluation that can be applied by ECLBS members. The model is normative rather than statistical. In other words, it is designed to help institutions make clear and fair decisions, even across different academic systems.
The proposed thresholds are intentionally simple. They are not presented as a universal law. They are presented as an institutional standard that can improve consistency and communication.
Analysis
1. Why clear thresholds matter
Students often receive unclear messages about plagiarism. One supervisor may say that any similarity is dangerous. Another may tolerate high percentages if the writing looks formal. This inconsistency creates confusion and increases disputes.
A threshold model solves part of this problem. It tells students from the beginning what range is generally safe, what range needs careful review, and what range is unacceptable. In quality assurance, clarity is itself a form of prevention.
2. Less than 10% = Acceptable
A similarity score below 10% is generally a reasonable sign that the thesis is mostly original in wording and structure. Small overlaps are normal in academic writing. Titles, common phrases, methodological language, and correctly cited short expressions may appear in many documents. Therefore, a score below 10% should normally be acceptable, provided that there is no sign of hidden misconduct.
This threshold is also practical. It encourages students to aim for strong originality without punishing normal academic conventions.
3. 10–15% = Needs Evaluation
The middle zone is the most important. A score between 10% and 15% should not be treated as automatic failure. Instead, it should trigger academic evaluation.
At this stage, the examiner should ask:
Is the overlap concentrated in one section or spread across the thesis?
Are the matched parts properly cited?
Is the issue limited to definitions and standard terminology?
Does the thesis show original argument, analysis, and interpretation?
Was AI used for editing only, or did it shape large parts of the written content?
This middle category protects fairness. It recognizes that numbers alone cannot judge intent or quality. It also gives institutions room to distinguish weak academic practice from serious misconduct.
4. Above 15% = Fail
A similarity score above 15% should normally lead to failure, especially in thesis work. At this level, the risk is too high that originality has been compromised. Even if some overlaps are technical, such a score calls into question whether the thesis represents independent academic effort.
A fail decision at this level also sends a clear quality message. If a thesis is the final proof of academic readiness, institutions must protect its credibility. Allowing high similarity too easily can damage the reputation of the qualification itself.
However, procedural fairness still matters. The student should be allowed to see the report and receive a formal explanation. In some systems, resubmission after major revision may be possible. But the original result should remain a fail if the threshold is exceeded and the review confirms unacceptable overlap.
5. AI thresholds and disclosure
Similarity percentages alone do not solve the AI question. AI-generated text may not always appear in similarity software. Therefore, institutions should add a second rule: any material use of AI in a thesis must be declared by the student.
This means students should disclose if AI was used for:
language polishing beyond minor grammar correction,
summarizing sources,
generating outlines,
suggesting text passages,
creating tables, arguments, or literature maps.
Undisclosed substantive AI use should be treated as an academic integrity concern even when similarity is low. A thesis can be original in software terms and still be academically weak if the student did not produce the intellectual work personally.
6. Why this is positive for education
A clear standard is not anti-student. It is pro-quality. It protects honest students, gives confidence to supervisors, and supports institutional trust. In a positive education culture, rules are not designed only to catch mistakes. They are designed to create confidence in achievement.
For ECLBS members, adopting a shared recommendation could also improve comparability across institutions. That is especially valuable in an international academic environment where students, employers, and quality reviewers expect transparency.
Findings
This article identifies six main findings.
First, current positive developments in Latvia’s education and academic governance create a timely context for stronger institutional attention to thesis quality and integrity.
Second, plagiarism policy must now include both traditional text copying and the responsible governance of AI-assisted writing.
Third, a three-level threshold model is easier to communicate and apply than vague ethical statements alone.
Fourth, Less than 10% = Acceptable is a reasonable baseline for originality.
Fifth, 10–15% = Needs Evaluation is necessary because academic judgment must remain part of the process.
Sixth, Above 15% = Fail is a defensible quality standard for theses, provided due process is respected.
These findings support a practical recommendation for ECLBS members: adopt the threshold model together with mandatory AI disclosure and supervisor guidance.
Conclusion
This week’s positive educational momentum in Latvia is a reminder that quality in education depends not only on reforms, structures, and competitiveness, but also on trust in academic work. Thesis evaluation is one of the most visible places where that trust is tested.
In the age of AI, academic integrity rules must become simpler, clearer, and more realistic. Institutions should not rely only on software or only on personal judgment. They need both. The proposed framework offers a balanced path:
Less than 10% = Acceptable
10–15% = Needs Evaluation
Above 15% = Fail
Alongside this, ECLBS members should require transparent disclosure of substantive AI use in thesis preparation. This approach is practical, fair, and aligned with a modern quality culture. It protects originality without rejecting innovation. It also supports a positive academic environment in which students know the rules, supervisors apply consistent standards, and institutions maintain the credibility of their awards.
For academic communities that aim to strengthen quality, this is not a minor administrative issue. It is a core educational responsibility.
References
Bretag, T. Handbook of Academic Integrity. Springer.
Carroll, J. A Handbook for Deterring Plagiarism in Higher Education. Oxford Centre for Staff and Learning Development.
Eaton, S. E. Plagiarism in Higher Education: Tackling Tough Topics in Academic Integrity. Libraries Unlimited.
Foltýnek, T., Meuschke, N., and Gipp, B. “Academic Plagiarism Detection: A Systematic Literature Review.” ACM Computing Surveys.
Pecorari, D. Academic Writing and Plagiarism: A Linguistic Analysis. Continuum.
Sowden, C. “Plagiarism and the Culture of Multilingual Students in Higher Education Abroad.” ELT Journal.
Sutherland-Smith, W. Plagiarism, the Internet, and Student Learning: Improving Academic Integrity. Routledge.
Williamson, B., Eynon, R., and Potter, J. “Pandora’s Box of Artificial Intelligence in Education.” Learning, Media and Technology.
Zhai, X., Chu, X., Chai, C. S., et al. “A Review of Artificial Intelligence in Education.” Computers and Education: Artificial Intelligence.
Comments