Doing a Research Thesis and AI

Graduate students increasingly utilize AI tools to enhance the efficiency and quality of their research theses. Here are ten notable AI tools commonly employed in academic research:

1. Grammarly: An AI-powered writing assistant that checks for grammar, punctuation, and style errors, ensuring clarity and correctness in academic writing. Grammarly
2. QuillBot: A paraphrasing tool that helps rephrase sentences and paragraphs, avoiding plagiarism and improving the readability of text. QuillBot
3. Zotero: A reference management tool that collects, organizes, cites, and shares research materials, streamlining the citation process. Zotero
4. Mendeley: A reference manager and academic social network that helps organize research, collaborate with others online, and discover the latest research. Mendeley
5. EndNote: A reference management software that allows users to manage bibliographies and references when writing essays and articles. EndNote
6. Turnitin: A plagiarism detection service that checks submitted documents for originality, ensuring academic integrity. Turnitin
7. Iris.ai: An AI research assistant that helps find relevant scientific papers, extract key data points, and facilitate literature reviews. Iris.ai
8. Scholarcy: An AI-powered article summarizer that creates summaries and flashcards from research papers, aiding in quick comprehension. Scholarcy
9. Writefull: An AI tool that provides language feedback tailored to academic writing, helping improve grammar, vocabulary, and style. Writefull
10. Elicit: An AI research assistant that helps in literature review by finding relevant papers and summarizing their key points. Elicit

These tools assist graduate students in various aspects of thesis writing, from ensuring proper grammar and style to managing references and detecting plagiarism, thereby enhancing the overall quality and efficiency of academic research.

In response to students' increasing use of generative AI tools, educators are adopting innovative assessment methods to evaluate graduate academic performance accurately. These strategies aim to minimize AI's influence and emphasize authentic student learning: 

1. Process-Oriented Assessments: Emphasizing the steps in completing assignments, such as research proposals, outlines, and drafts, allows instructors to monitor student progress and understanding. This approach makes it challenging for students to rely solely on AI-generated content.
   GenAI
2. Personalized and Contextual Tasks: Assignments tailored to individual experiences or current events require students to apply knowledge uniquely, reducing the effectiveness of generic AI responses.
   Northern Michigan University
3. Reflective Components: Incorporating reflective elements encourages students to articulate their thought processes, decision-making, and learning experiences, which are difficult for AI to replicate authentically.
   GenAI
4. In-Person Presentations and Oral Exams: Assessments involving real-time demonstrations of knowledge, such as presentations or oral exams, ensure that students comprehend the material and can communicate it effectively without AI assistance.
   Concordia University Alberta
5. Collaborative Projects: Group assignments promote teamwork and require students to engage actively with peers, making it harder to depend on AI tools.
   Times Higher Education
6. Authentic Assessments: Designing tasks that mirror real-world scenarios and require practical application of skills ensures that students demonstrate genuine understanding and abilities.
   Central Queensland University
7. Adaptive Comparative Judgement: This method involves comparing pairs of student work to establish a rank order based on quality. It focuses on holistic judgment rather than specific criteria, which AI may not effectively influence.
   Wikipedia

By integrating these strategies, educators can create assessment environments that prioritize critical thinking, originality, and authentic demonstration of knowledge, thereby minimizing the impact of AI support in student evaluations.

Tutors use various strategies and tools to assess AI involvement in a student's research thesis, focusing on detecting patterns, verifying originality, and encouraging transparent usage. Here are some common methods:

1. Plagiarism and Similarity Detection Tools: Traditional plagiarism tools like Turnitin now offer AI detection features that can identify AI-generated text by analyzing language patterns, sentence structure, and other unique markers of AI-generated content.
2. Comparing Writing Styles: Tutors often become familiar with each student's unique writing style over time. Sudden changes in vocabulary, tone, or sentence complexity might indicate the use of AI, prompting a closer review.
3. Requesting Process Documentation: Tutors may ask students to submit drafts, outlines, and progress updates, which helps them track the development of ideas and identify inconsistencies that might suggest AI assistance.
4. Focused Oral Examinations: Tutors may conduct oral exams or presentations on specific thesis sections to confirm understanding. In these sessions, students are asked to explain methodologies, data interpretations, and theoretical frameworks. Difficulty in providing clear, consistent answers might reveal heavy AI involvement.
5. Annotation Requirements: Tutors sometimes ask students to annotate specific sections of their thesis, detailing how they arrived at certain conclusions or processed data. This requires students to go beyond surface-level knowledge, discouraging AI reliance.
6. Use of Specialized AI Detection Software: Some universities and academic institutions invest in advanced AI detection tools like GPTZero or Originality.ai, which analyze text patterns and flag content likely generated by AI models.
7. Inquiry into Data and Methodology Sources: Tutors can ask students to detail the specific data sources, tools, and techniques used in their research. If students cannot explain the rationale or process behind these, it may suggest AI was used in place of genuine research.
8. Cross-Referencing Sources: Tutors may verify that cited sources and references align logically with the content. AI-generated work sometimes includes fictitious or poorly matched references, so checking citations is another method for detecting AI assistance.

By combining these approaches, tutors can better assess a student's contribution and ensure the authenticity of thesis work, helping maintain academic integrity in an AI-influenced environment.