مراجعة منهجية شاملة للذكاء الاصطناعي في التعليم العالي: دعوة لزيادة الأخلاقيات، والتعاون، والصرامة
A meta systematic review of artificial intelligence in higher education: a call for increased ethics, collaboration, and rigour

المجلة: International Journal of Educational Technology in Higher Education، المجلد: 21، العدد: 1
DOI: https://doi.org/10.1186/s41239-023-00436-z
تاريخ النشر: 2024-01-19
المؤلف: Melissa Bond وآخرون
الموضوع الرئيسي: التعلم عبر الإنترنت والتحليلات

نظرة عامة

تقدم هذه القسم نظرة عامة على مراجعة شاملة تركز على الذكاء الاصطناعي في التعليم (AIEd) ضمن التعليم العالي، والتي تُعرف بـ AIHEd. تقوم هذه المراجعة بتلخيص الأبحاث الثانوية المنشورة بين عامي 2018 ويوليو 2023، مستندة إلى مجموعة متنوعة من قواعد البيانات الأكاديمية والمنصات. تم تضمين ما مجموعه 66 منشورًا، معظمها مراجعات منهجية، مع تركيز ملحوظ على مواضيع مثل التطبيقات العامة لـ AIHEd والتوصيف والتنبؤ. تكشف النتائج عن تركيز كبير على الأنظمة التكيفية والتخصيص في التعليم العالي، مما يبرز الاتجاهات الحالية في أبحاث AIEd.

تؤكد الخاتمة على أهمية هذه المراجعة الثانوية في رسم خريطة مشهد AIHEd، وفحص أنماط التأليف، وجودة البحث، وتحديد الموضوعات الرئيسية والفجوات في الأدبيات. كما تؤكد على ضرورة أن تتناول الأبحاث المستقبلية الاعتبارات الأخلاقية والمنهجية والسياقية، داعية إلى اتباع نهج متعدد التخصصات لتعزيز تطبيق الذكاء الاصطناعي في التعليم. تضع المراجعة الأساس لاستكشاف أعمق لمجموعة أوسع من 307 تجميعات أدلة AIEd، بهدف تقديم رؤى أعمق وإرشادات لمشاريع البحث المستقبلية في هذا المجال.

مقدمة

تسلط مقدمة ورقة البحث الضوء على تطور الذكاء الاصطناعي (AI) في التعليم، متتبعة جذوره إلى الستينيات مع ظهور أنظمة التدريس الذكية. على الرغم من الاعتراف المتزايد بإمكانات الذكاء الاصطناعي في البيئات التعليمية، إلا أن تطبيقه العملي في الفصول الدراسية لم يكتسب الزخم إلا مؤخرًا، خاصة مع ظهور أدوات الذكاء الاصطناعي التوليدية مثل ChatGPT وDALL-E. أثارت هذه التطورات مناقشات كبيرة حول تداعيات الذكاء الاصطناعي في التعليم، بما في ذلك مواضيع الاستعداد والأخلاقيات والثقة وضرورة الحوكمة والتدريب للتنقل في التحول السريع في التعليم والتعلم.

استجابةً للتأثير المتزايد للذكاء الاصطناعي، تقوم حكومات مختلفة بتنفيذ تدابير تنظيمية، مثل قانون الذكاء الاصطناعي في الاتحاد الأوروبي في أوروبا ومبادرات في أستراليا والولايات المتحدة تهدف إلى إنشاء أطر لدمج الذكاء الاصطناعي في السياقات التعليمية. يؤكد المؤلفون على أهمية تأسيس هذه الإجراءات التنظيمية على أبحاث قوية، مشيرين إلى أنه على الرغم من وجود مجتمع نابض يركز على الذكاء الاصطناعي في التعليم (AIEd)، لا يزال هناك الكثير من العمل الأساسي الذي يتعين تطويره. تقدم هذه الورقة مراجعة ثانوية، مما يمثل جهدًا رائدًا في أبحاث AIEd، والتي تهدف إلى وضع الأساس للتصور والتطبيق المستقبلي للذكاء الاصطناعي في التعليم العالي.

الطرق

في قسم الطرق، يناقش المؤلفون تطور تقنيات تجميع الأدلة، لا سيما ضمن مجال الرعاية الصحية. يبرزون أن مراجعات الأدبيات، التي كانت تقليديًا الشكل الأكثر اعترافًا به من الأبحاث الثانوية، قد تنوعت إلى أنواع مختلفة من المراجعات. حدد سوتون وآخرون (2019) 48 نوعًا متميزًا من المراجعات، مصنفة إلى سبع عائلات، بما في ذلك الطرق المختلطة ومراجعة المراجعات. تؤكد هذه التصنيف على التعقيد المتزايد والطبيعة المنهجية لمراجعات الأدبيات المعاصرة، التي غالبًا ما تتضمن أقسامًا منهجية مع الاحتفاظ بالتصنيف التقليدي.

بالإضافة إلى ذلك، يشير المؤلفون إلى ارتفاع تحليلات الببليومترية كنوع مهم من تجميع الأدلة. تستخدم هذه التحليلات البيانات الببليوغرافية لتقييم اتجاهات البحث وتأثيراته، مما يوفر رؤى قيمة في مجالات محددة. على الرغم من عدم تضمينها في إطار سوتون وآخرون، يجادل المؤلفون من أجل الاعتراف بها كطريقة مهمة لتجميع الأدلة نظرًا لقدرتها على إلقاء الضوء على ديناميات البحث.

النتائج

يقدم قسم ورقة البحث النتائج الرئيسية من تجميع الأدلة المتعلقة بتطبيق الذكاء الاصطناعي في التعليم العالي، مؤطرًا ضمن نموذج دورة حياة الطالب (ريد، 1995). تكشف التحليلات عن تركيز رئيسي على خدمات الدعم الأكاديمي، حيث تناولت 97% (n = 64) من الدراسات المراجعة التعليم والتعلم، بينما فقط 39.3% (n = 26) فحصت الخدمات المؤسسية والإدارية. يتماشى هذا الاتجاه مع النتائج السابقة من كرومبتون وبيرك (2023)، التي أشارت إلى أن 11% فقط من أبحاث التعليم العالي تركزت على الجوانب الإدارية، على الرغم من إمكانية الذكاء الاصطناعي لتعزيز التخصيص في القبول، والامتحانات، وخدمات المكتبة (الجابري وآخرون، 2021؛ زواكي-ريختر وآخرون، 2019).

حددت التصنيف الموضوعي لتطبيقات الذكاء الاصطناعي أربعة مجالات رئيسية: الأنظمة التكيفية والتخصيص (54.5%، n = 36)، التوصيف والتنبؤ (48.5%، n = 32)، التقييم والتقويم (39.4%، n = 26)، وأنظمة التدريس الذكية (21.2%، n = 14). تؤكد هذه النتائج على التركيز الكبير على تقنيات التعلم التكيفية، بينما تبرز إهمالًا نسبيًا لأنظمة التدريس الذكية ضمن الجسم الحالي من الأبحاث. يوفر تجميع هذه النتائج نظرة شاملة على المشهد الحالي لتطبيقات الذكاء الاصطناعي في التعليم العالي، مما يشير إلى مجالات للنمو المحتمل والتحقيق الإضافي.

المناقشة

تسلط قسم المناقشة في هذه المراجعة الضوء على أهمية تجميع الأدلة في مجال الذكاء الاصطناعي في التعليم (AIEd) للتخفيف من خطر “إهدار البحث” الناتج عن الجهود المكررة. من خلال اعتماد نهج مراجعة ثانوية، تهدف هذه الدراسة إلى تقديم نظرة شاملة على الحالة الحالية لأبحاث AIEd في التعليم العالي (AIHEd). يجمع هذا الأسلوب النتائج من الدراسات الثانوية الموجودة، مما يسمح بتحديد الخصائص الرئيسية، والفجوات في الأدبيات، وتوجيه اتجاهات البحث المستقبلية. تعتبر المراجعة ملائمة بشكل خاص نظرًا للاهتمام المتزايد في تطبيقات الذكاء الاصطناعي التوليدية في السياقات التعليمية.

تركز أسئلة البحث التي توجه هذه المراجعة على طبيعة ونطاق تجميع الأدلة في AIHEd، بما في ذلك أنواع التجميعات المنجزة، ووسائل النشر، وتوزيع التأليف الجغرافي، ومستويات التعاون، وجودة الأدلة المجمعة. تهدف المراجعة أيضًا إلى استكشاف التطبيقات الرئيسية لـ AIEd والفوائد والتحديات المبلغ عنها في الأدبيات. من خلال رسم خريطة منهجية لهذه الجوانب، تسعى المراجعة إلى إبلاغ أصحاب المصلحة حول التقدم والتحديات المستمرة في دمج تقنيات الذكاء الاصطناعي في الممارسات التعليمية، مما يسهم في تطوير هذا المجال.

القيود

في هذا القسم، يعترف المؤلفون بعدة قيود في مراجعتهم الميتا. من الجدير بالذكر أن بروتوكول المراجعة لم يتم تسجيله مسبقًا في مستودع رسمي مثل بروسبيرو، حيث إنها مراجعة ثانوية وليست دراسة طبية أولية. ومع ذلك، جعلوا جميع معلومات البحث متاحة للجمهور عبر إطار العلوم المفتوحة (OSF) ويخططون لاستخدام منظمات معروفة للمراجعات المستقبلية. بالإضافة إلى ذلك، قيد المؤلفون بحثهم على أول 500 سجل في جوجل سكولار، وهو أقل من 1000 سجل الموصى بها، على الرغم من أنهم عوضوا عن ذلك باستخدام OpenAlex. كما يعترفون بأن المزيد من عمليات البحث اليدوية في مجلات أكاديمية محددة والأدبيات بلغات أخرى غير الإنجليزية كان يمكن أن يقلل من التحيز اللغوي المحتمل.

علاوة على ذلك، يشير المؤلفون إلى أن أداة تقييم الجودة التي طوروها لها قيودها، لا سيما فيما يتعلق بالافتراض بوجود مسافة متساوية بين فئات الاستجابة (نعم، لا، جزئيًا). على الرغم من هذه المخاوف، يشيرون إلى أن الأدوات المستخدمة معترف بها على نطاق واسع وقد تم استخدامها في أبحاث سابقة، مما يعزز من صحة نهجهم المنهجي.

Journal: International Journal of Educational Technology in Higher Education, Volume: 21, Issue: 1
DOI: https://doi.org/10.1186/s41239-023-00436-z
Publication Date: 2024-01-19
Author(s): Melissa Bond et al.
Primary Topic: Online Learning and Analytics

Overview

The section provides an overview of a comprehensive meta-review focused on Artificial Intelligence in Education (AIEd) within higher education, termed AIHEd. This review synthesizes secondary research published between 2018 and July 2023, drawing from various academic databases and platforms. A total of 66 publications were included, primarily systematic reviews, with a notable concentration on themes such as general AIHEd applications and Profiling and Prediction. The findings reveal a significant emphasis on Adaptive Systems and Personalisation in higher education, highlighting the current trends in AIEd research.

The conclusion underscores the importance of this tertiary review in mapping the AIHEd landscape, examining authorship patterns, research quality, and identifying key topics and gaps in the literature. It emphasizes the necessity for future research to address ethical, methodological, and contextual considerations, advocating for interdisciplinary approaches to enhance the application of AI in education. The review sets the stage for further exploration of the broader corpus of 307 AIEd evidence syntheses, aiming to provide deeper insights and guidance for future research endeavors in the field.

Introduction

The introduction of the research paper highlights the evolution of Artificial Intelligence (AI) in education, tracing its roots back to the 1960s with the emergence of intelligent tutoring systems. Despite the increasing recognition of AI’s potential in educational settings, its practical application in classrooms has only recently gained momentum, particularly with the advent of generative AI tools like ChatGPT and DALL-E. These developments have sparked significant discussions regarding the implications of AI in education, encompassing themes of readiness, ethics, trust, and the necessity for governance and training to navigate the rapid transformation of teaching and learning.

In response to the growing influence of AI, various governments are implementing regulatory measures, such as the EU AI Act in Europe and initiatives in Australia and the United States aimed at establishing frameworks for AI integration in educational contexts. The authors emphasize the importance of grounding these regulatory actions in robust research, noting that while there exists a vibrant community focused on AI in education (AIEd), much foundational work remains to be developed. This paper presents a tertiary review, marking a pioneering effort in AIEd research, which aims to lay the groundwork for future conceptualization and application of AI in higher education.

Methods

In the section on methods, the authors discuss the evolution of evidence synthesis techniques, particularly within the healthcare domain. They highlight that literature reviews, traditionally the most recognized form of secondary research, have diversified into various types of reviews. Sutton et al. (2019) identified 48 distinct review types, categorized into seven families, including mixed methods and review of reviews. This classification underscores the increasing complexity and systematic nature of contemporary literature reviews, which often incorporate methodological sections while retaining the traditional label.

Additionally, the authors note the rise of bibliometric analyses as a significant form of evidence synthesis. These analyses utilize bibliographic data to assess research trends and impacts, providing valuable insights into specific fields. Although not included in Sutton et al.’s framework, the authors argue for their recognition as an important method of evidence synthesis due to their capacity to illuminate research dynamics.

Results

The research paper section presents key findings from a synthesis of evidence regarding the application of Artificial Intelligence in higher education, framed within the student life-cycle model (Reid, 1995). The analysis reveals a predominant focus on academic support services, with 97% (n = 64) of the reviewed studies addressing teaching and learning, while only 39.3% (n = 26) examined institutional and administrative services. This trend aligns with previous findings by Crompton and Burke (2023), which indicated that merely 11% of higher education research concentrated on managerial aspects, despite the potential of AI to enhance personalization in admissions, examinations, and library services (Algabri et al., 2021; Zawacki-Richter et al., 2019).

The thematic classification of AI applications identified four main areas: adaptive systems and personalization (54.5%, n = 36), profiling and prediction (48.5%, n = 32), assessment and evaluation (39.4%, n = 26), and intelligent tutoring systems (21.2%, n = 14). These findings underscore the significant emphasis on adaptive learning technologies, while highlighting a relative neglect of intelligent tutoring systems within the current body of research. The synthesis of these results provides a comprehensive overview of the current landscape of AI applications in higher education, indicating areas for potential growth and further investigation.

Discussion

The discussion section of this review highlights the importance of synthesizing evidence in the field of Artificial Intelligence in Education (AIEd) to mitigate the risk of ‘research waste’ caused by duplicated efforts. By employing a tertiary review approach, this study aims to provide a comprehensive overview of the current state of AIEd research in higher education (AIHEd). This method synthesizes findings from existing secondary studies, allowing for the identification of key characteristics, gaps in the literature, and guiding future research directions. The review is particularly timely given the growing interest in generative AI applications in educational contexts.

The research questions guiding this review focus on the nature and scope of evidence synthesis in AIHEd, including the types of syntheses conducted, publication venues, geographical authorship distribution, collaboration levels, and the quality of the evidence synthesized. The review also aims to explore the main applications of AIEd and the benefits and challenges reported in the literature. By systematically mapping these aspects, the review seeks to inform stakeholders about the advancements and ongoing challenges in integrating AI technologies into educational practices, ultimately contributing to the development of the field.

Limitations

In this section, the authors acknowledge several limitations of their meta review. Notably, the review protocol was not pre-registered in an official repository like Prospero, as it is a tertiary review rather than a primary medical study. However, they have made all search information publicly accessible via the Open Science Framework (OSF) and plan to utilize established organizations for future reviews. Additionally, the authors limited their search to the first 500 records in Google Scholar, which is below the recommended 1000 records, although they supplemented this with OpenAlex. They also recognize that further manual searches of specific academic journals and literature in languages other than English could have mitigated potential language bias.

Moreover, the authors point out that the quality assessment tool they developed has its limitations, particularly regarding the assumption of equal distance between response categories (yes, no, partly). Despite these concerns, they note that the tools employed are widely recognized and have been utilized in previous research, reinforcing the validity of their methodological approach.