ما هي محو الأمية والكفاءة في الذكاء الاصطناعي؟ إطار شامل لدعمهما What are artificial intelligence literacy and competency? A comprehensive framework to support them

المجلة: Computers and Education Open، المجلد: 6
DOI: https://doi.org/10.1016/j.caeo.2024.100171
تاريخ النشر: 2024-03-13

ما هي محو الأمية والكفاءة في الذكاء الاصطناعي؟ إطار شامل لدعمهما

توماس ك.ف. تشيو زبير أحمد مورد إسماعيلوف إسماعيل تميتايو سانوسي قسم المناهج والتدريس، مركز علوم التعلم والتكنولوجيا ومركز الشراكة بين الجامعة والمدرسة في الجامعة الصينية في هونغ كونغ تطوير ونشر الوحدات، جامعة قطر معهد العلوم الإنسانية والاجتماعية، جامعة تسوكوبا كلية الحوسبة، جامعة فنلندا الشرقية مركز العلماء الشباب بجامعة قطر (QUYSC)، 2713، جامعة قطر، الدوحة، قطر

معلومات المقال

الكلمات المفتاحية:

محو الأمية في الذكاء الاصطناعي
كفاءة الذكاء الاصطناعي
التعليم من رياض الأطفال حتى الصف الثاني عشر
تعلم الآلة
محو الأمية البيانات
الذكاء الاصطناعي التوليدي

الملخص

تعليم الذكاء الاصطناعي في المدارس من رياض الأطفال حتى الصف الثاني عشر هو مبادرة عالمية، ومع ذلك فإن التخطيط وتنفيذ تعليم الذكاء الاصطناعي يمثل تحديًا. تركز الأطر الرئيسية على تحديد المحتوى والمعرفة التقنية (محو الأمية في الذكاء الاصطناعي). تم تطوير معظم التعريفات الحالية لمحو الأمية في الذكاء الاصطناعي لجمهور غير تقني من منظور هندسي وقد لا تكون مناسبة لـ التعليم. تعتبر وجهات نظر المعلمين ضرورية لفهم هذه المبادرة. تتعلق محو الأمية بمعرفة (المعرفة، ما المهارات)؛ بينما تتعلق الكفاءة بتطبيق المعرفة بطريقة مفيدة (الثقة، مدى الكفاءة). وهما مرتبطتان ارتباطًا وثيقًا. تتجاوز هذه الدراسة المعرفة (محو الأمية في الذكاء الاصطناعي)، وتهدف إلى (i) تعريف محو الأمية والكفاءة في الذكاء الاصطناعي من خلال إضافة جوانب الثقة والعقلية الانعكاسية الذاتية، و(ii) اقتراح إطار عمل أكثر شمولاً لتعليم الذكاء الاصطناعي في المرحلة K-12. هذه التعريفات ضرورية لهذه التكنولوجيا الناشئة والم disruptive (مثل ChatGPT وSora، الذكاء الاصطناعي التوليدي). استخدمنا التعريفات ونماذج تصميم المناهج الأساسية كإطار تحليلي ووجهات نظر المعلمين. شمل المشاركون 30 معلمًا ذو خبرة في الذكاء الاصطناعي من 15 مدرسة متوسطة. استخدمنا دورة تصميم مشتركة تكرارية لمناقشة وتنقيح الإطار على مدار أربع دورات. يتضمن تعريف كفاءة الذكاء الاصطناعي خمس قدرات تأخذ في الاعتبار الثقة، ويتكون الإطار المقترح من خمسة مكونات رئيسية: التكنولوجيا، التأثير، الأخلاق، التعاون، والانعاكاس الذاتي. كما نحدد خمس تجارب تعلم فعالة لتعزيز القدرات والثقة، ونقترح خمسة اتجاهات بحث مستقبلية: هندسة المطالبات، محو الأمية في البيانات، محو الأمية الخوارزمية، العقلية الانعكاسية الذاتية، والبحث التجريبي.

مقدمة

الذكاء الاصطناعي (AI) هو قدرة آلة رقمية على تنفيذ مهام عادة ما يقوم بها الكائنات الذكية. تشمل تقنياته التي تدعم الذكاء الاصطناعي رؤية الكمبيوتر، وتحويل الكلام إلى نص، ومعالجة اللغة الطبيعية [7]. إن التقدم في الذكاء الاصطناعي له تأثيرات عميقة على حياتنا اليومية، والترفيه، والتعليم، والوظائف. من الضروري توسيع تدريب الذكاء الاصطناعي ليشمل ما هو أبعد من التعليم العالي والمحترفين. بهدف إعداد جميع المواطنين لمجتمع قائم على الذكاء الاصطناعي، تم تضمين تعليم الذكاء الاصطناعي في المجتمع غير المتخصص حول العالم، مثل: الذكاء الاصطناعي للجميع [24] والذكاء الاصطناعي للمرحلة الابتدائية والثانوية [4،43]. نحتاج جميع أطفالنا الصغار إلى أن يكون لديهم معرفة وكفاءة جيدة في الذكاء الاصطناعي [3]. لذلك، فإن تعليم الذكاء الاصطناعي لـ هي مبادرة عالمية، كما يتضح في منظمة اليونسكو
تقرير عن تعليم الذكاء الاصطناعي. من ناحية أخرى، على عكس التعليم العالي، تصميم يجب أن تأخذ التعليم في الاعتبار التنفيذ والتنوع في طرق التوصيل. لمعالجة المبادرة العالمية، تم اقتراح بعض الأطر الرئيسية للباحثين والمعلمين. تركز هذه الأطر على تحديد المحتوى والمعرفة لتعليم وتعلم الذكاء الاصطناعي. إنها مهمة جدًا في بداية هذه المبادرة لأن الباحثين والمعلمين لم يعرفوا ما يجب تضمينه في منهج الذكاء الاصطناعي. ومع ذلك، فإن تعليم الذكاء الاصطناعي يتجاوز المحتوى، لأنه يتناول نتائج التعلم (أي، ما هي الكفاءة والمعرفة في الذكاء الاصطناعي؟) والتجارب (أي، كيف يمكن تنميتها؟).
لقد عرفت الأدبيات واقترحت ما هي معرفة الذكاء الاصطناعي للمهنيين غير المتخصصين في الذكاء الاصطناعي (على سبيل المثال، [29،31]). قد لا يستفيد الأطفال الأصغر سناً من التعريف الذي تم تطويره من منظور هندسي. طويل
وذكر مارغيركو [31] أن تعريفهم يحتاج إلى مراجعة في المستقبل حيث أن البحث في تعليم الذكاء الاصطناعي لا يزال في مراحله المبكرة. التقنيات ناشئة وم disruptive، مثل ChatGPT و Sora. هذه الأدوات الذكية التوليدية تبقي البشر في حالة تعلم [9]. لذلك، يجادل هذا البحث بأن (i) يجب أن تشمل مفاهيم العقليات الذاتية الانعكاسية في تعريف معرفة الذكاء الاصطناعي (أي، كفاءة الذكاء الاصطناعي)، ويجب أخذ وجهات نظر الممارسين في الاعتبار عند تصميم وتطوير تعليم الذكاء الاصطناعي في المرحلة K-12. علاوة على ذلك، تصف نهج تصميم المناهج الأساسية – العملية والممارسة – كيفية تنمية معرفة الطلاب [20]. وبناءً عليه، فإن الهدف الرئيسي من هذه الدراسة هو اقتراح إطار عمل لـ المعلمين لتعزيز معرفة الذكاء الاصطناعي والكفاءة. يشرح هذا الإطار الشامل المعرفة الأساسية (المحتوى) وطرق التدريس (عملية التعلم) التي تساعد الطلاب على تحسين معرفتهم وكفاءتهم في الذكاء الاصطناعي (نتائج التعلم). من أجل تحقيق هذا الهدف، قمنا بتعريف معرفة الذكاء الاصطناعي والكفاءة وصممنا الإطار بالتعاون مع المعلمين. نظرًا لأن احتياجات التعلم للبالغين الذين لم يتلقوا تعليمًا في الذكاء الاصطناعي قابلة للمقارنة مع تلك الخاصة بـ يمكن تطبيق الإطار على كل من غير المتخصصين في الذكاء الاصطناعي وطلاب K-12، أي الذكاء الاصطناعي للجميع والذكاء الاصطناعي لطلاب K-12. ستساهم نتائج هذه الدراسة في تعليم الذكاء الاصطناعي من خلال تعزيز معرفتنا بالقدرة على فهم الذكاء الاصطناعي والكفاءة لطلاب K-12. سيستخدم الباحثون والممارسون التعريف والإطار النظري المقترح من قبل النتائج لتصميم أبحاث وأنشطة تعليمية تتعلق بفهم الذكاء الاصطناعي والكفاءة.

مراجعة الأدبيات

يتكون الإطار المفاهيمي لهذه الدراسة من تعريفنا لمحو الأمية في الذكاء الاصطناعي والكفاءة، بالإضافة إلى نهجين لتصميم المناهج: الممارسة والعملية. يقدم التعريف اقتراحات للمحتوى ونتائج التعلم التي يجب تضمينها في تعليم الذكاء الاصطناعي في المرحلة K-12؛ وتوصي النهجان باستخدام أساليب تدريس فعالة. تم تعريف مصطلحي محو الأمية في الذكاء الاصطناعي والكفاءة في هذا القسم، إلى جانب مناقشة النهجين لتصميم المناهج – العملية والممارسة. بعد ذلك تأتي المناقشة النقدية للأوراق ذات الصلة بإطار تعليم الذكاء الاصطناعي من أجل تحديد فجوات البحث.

محو الأمية والمهارة في الذكاء الاصطناعي

في البداية، تشير محو الأمية إلى طرق محددة للتفكير في القراءة والكتابة وأدائها من أجل فهم أو التعبير عن الأفكار أو الآراء كتابيًا ضمن سياق معين للاستخدام. تشير محو الأمية الرقمية إلى القدرة على استخدام وتقييم وتطبيق الأدوات والموارد والخدمات الرقمية بشكل مناسب في عمليات التعلم مدى الحياة. من الواضح أن الكفاءة الرقمية تشمل أكثر من مجرد الكفاءة في تشغيل الأجهزة والبرامج؛ فهي مرتبطة أيضًا ارتباطًا وثيقًا بالقدرة على التواصل باستخدام التقنيات والمهارات الرقمية. يجب أن تتضمن رؤية متوازنة للتكنولوجيا للاستخدام المسؤول والصحي للتكنولوجيا الرقمية. يجب أن تشمل المعرفة والمواقف حول الخصوصية والأمان، والاعتبارات القانونية والأخلاقية، ودور التقنيات الرقمية في المجتمع. لذلك، فإن محو الأمية مرتبط مباشرة بالمهارات ويدور أكثر حول المعرفة. بخلاف ذلك، فإن الكفاءة هي القدرة على أداء مهمة بشكل فعال وناجح. بشكل عام، تدور محو الأمية حول المعرفة، بينما تدور الكفاءة حول تطبيق المعرفة بطريقة فعالة ومفيدة. وهي مرتبطة بالثقة والموقف، وتركز على مدى جودة أداء مستخدم الذكاء الاصطناعي.
تعرف لونغ وماجيركو [31] محو الأمية في الذكاء الاصطناعي بأنه “مجموعة من الكفاءات التي تمكن الأفراد من تقييم تقنيات الذكاء الاصطناعي بشكل نقدي، والتواصل والتعاون بفعالية مع الذكاء الاصطناعي، واستخدام الذكاء الاصطناعي كأداة عبر الإنترنت، في المنزل، وفي مكان العمل.” يرون أن هذه الأمية هي مجموعة من 17 مهارة وتعريف تشغيلي. من الواضح أن محو الأمية في الذكاء الاصطناعي مرتبط بأشكال أخرى من محو الأمية مثل محو الأمية الرقمية، ومحو الأمية البيانية، ومحو الأمية الحاسوبية. يمكن أن تكون العلاقات متبادلة الاعتماد ولكنها حصرية [31]. على سبيل المثال، تتطلب محو الأمية في الذكاء الاصطناعي من المستخدمين أن يكون لديهم فهم أساسي لكيفية استخدام الحواسيب من أجل فهم الذكاء الاصطناعي. لذلك، فإن محو الأمية الرقمية، التي تشير إلى القدرة على استخدام الحواسيب لإكمال مهمة [33]، هي
شرط أساسي لمحو الأمية في الذكاء الاصطناعي. بالنظر إلى العلاقة الوثيقة بين البيانات وتعلم الآلة (فرع من الذكاء الاصطناعي)، تشير محو الأمية البيانية إلى القدرة على فهم، والعمل مع، وتقييم، والجدل حول البيانات كجزء من عملية استقصاء أكثر شمولاً حول العالم [45]، والتي تتداخل إلى حد كبير مع محو الأمية في الذكاء الاصطناعي. علاوة على ذلك، قد لا ترتبط محو الأمية الأخرى، مثل الحاسوبية والعلمية، ارتباطًا وثيقًا بمحو الأمية في الذكاء الاصطناعي. تتضمن محو الأمية الحاسوبية استكشاف وتواصل الأفكار من خلال الشيفرة [44]؛ لذلك، ليست شرطًا أساسيًا لمحو الأمية في الذكاء الاصطناعي التي لا تتطلب كتابة الشيفرات لفهم كيفية عمل الذكاء الاصطناعي. وبالمثل، لا تتطلب محو الأمية في الذكاء الاصطناعي محو الأمية العلمية، التي تشير إلى تقدير طبيعة العلم، ومساهماته، والقيود الأساسية له [22]. تعريف محو الأمية في الذكاء الاصطناعي هو واحد من الأولويات للمحترفين غير المتخصصين في الذكاء الاصطناعي، مما قد يوفر للمعلمين والباحثين رؤى جديدة حول المحتوى وتقييم المهارات المتعلقة بالذكاء الاصطناعي. تم تطويره بناءً على مراجعة أدبية أجراها أستاذان في الهندسة. تم نشر معظم الأدبيات في مؤتمرات الهندسة حول عامي 2018-2019. وهذا يعني أن هذا التعريف مدفوع بمنظور هندسي والتعليم العالي. قد لا يكون التعريف مناسبًا لـ التعليم، وهو ما تدعمه القيود الرئيسية لهذا التعريف التي أشار إليها المؤلفون [31]. لقد لاحظوا أن تعليم الذكاء الاصطناعي لا يزال في مراحله الأولى ويتطلب دراسات تجريبية إضافية، خاصة حول وجهات نظر المعلمين، للحصول على فهم قوي ودقيق لمحو الأمية في الذكاء الاصطناعي لجمهور غير تقني [31]. علاوة على ذلك، في التعريف، تم استخدام مصطلحات محو الأمية والكفاءة بشكل مختلط في الأدبيات (على سبيل المثال، [31]). ومع ذلك، كما ناقشنا، فإن محو الأمية والكفاءة مرتبطتان ولكن مختلفتان في البحث التعليمي [17]. لذلك، من الضروري إعادة تعريف محو الأمية والكفاءة في الذكاء الاصطناعي.

طرق تصميم العمليات والممارسات في سياق تعليم الذكاء الاصطناعي

إن اعتماد استراتيجية تصميم منهجية محددة له تأثير كبير على عمليات التعليم والتعلم [38]. approaches الأربعة الأساسية لتصميم المناهج هي المحتوى، والمنتج، والعملية، والممارسة. تعتبر طريقة المحتوى التعليم كعملية نقل المعرفة، على سبيل المثال، منهج دراسي بمحتوى محدد [20،27]؛ تعتبر منهجية المنتج التعليم كعملية تقييم وتركز على أداء الطلاب [20،27]، أي نتائج تعلم الطلاب. غالبًا ما تنشئ قوائم من الكفاءات، تخبر الطلاب بما يجب عليهم تعلمه وكيفية تعلمه، وتؤكد على التعليم مع نتائج محددة مسبقًا. في تعليم الذكاء الاصطناعي، يتم إبلاغ هذين النهجين من خلال تعريف محو الأمية والكفاءة في الذكاء الاصطناعي.
بدلاً من المحتوى والنتائج المحددة مسبقًا، تؤكد طريقة تصميم العملية على كيفية تكيف المعلمين والطلاب والمحتوى، كما أنها تعتبر التعليم كعملية تطوير. تغير التفاعلات الثلاثية أهداف التعلم [27]. المنهج هو دليل للتعليم، وليس مجموعة من العناصر التي يجب على المعلمين تغطيتها وتقديمها [20]. يتم تخصيص المحتوى لاحتياجات واهتمامات الطلاب؛ تحدد نتائج التعلم من قبل المعلمين والطلاب ولكنها ليست قابلة للتطبيق عالميًا. تقدر هذه الطريقة تجارب التعلم المتمحورة حول الطالب. لفهم ما يتم تعلمه، تركز طريقة تصميم الممارسة على كيفية استخدامه في العالم الحقيقي. يتعاون الطلاب، تحت إشراف معلميهم، في إيجاد حلول لمشاكل العالم الحقيقي ويطورون استراتيجية لتعلم المواد اللازمة والحصول على النتائج المرغوبة. يتم تقييم كل من عملية التعلم ونتائجها باستمرار. لذلك، غالبًا ما يتم اعتماد التعلم القائم على المشكلات. لا يزال تعليم الذكاء الاصطناعي جديدًا على المدارس؛ إن طرقه وممارساته ليست واضحة وتتطلب المزيد من البحث التجريبي.
بشكل عام، تخلق الطريقتان الأوليان مجموعة من الوثائق للتنفيذ، وتدعو الطريقتان الأخيرتان إلى نهج متمحور حول الطالب، مما يحول التركيز من التعليم إلى التعلم [27]. لذلك، ستساهم تعريفات محو الأمية والكفاءة في الذكاء الاصطناعي ووجهات نظر المعلمين حول أساليب التعليم في تطوير إطار تعليم الذكاء الاصطناعي للمرحلة K-12.

ثلاثة أطر رئيسية لتعليم الذكاء الاصطناعي في K-12

تم اقتراح بعض الأطر الرئيسية في تعليم الذكاء الاصطناعي K-12
في السنوات الخمس الماضية. كان أحد الأطر الأولى، المعروف باسم “خمسة أفكار كبيرة في الذكاء الاصطناعي”، قد اقترحه تويريتسكي وآخرون [43]. في عام 2018، كان هناك القليل من التوجيه الخارجي من الأدبيات حول محتوى ونطاق تعليم الذكاء الاصطناعي لطلاب K-12 [42]. بدأت لجنة توجيه AI4K12، التي تتكون من ديفيد تويريتسكي، كريستينا غاردنر-ماكون، فريد مارتن، وديبورا سيهورن، عملها من خلال وضع قائمة. تعمل هذه القائمة كإطار تنظيمي للإرشادات، التي تم تطويرها بناءً على معايير CSTA للحوسبة. تم هيكلة تلك المعايير حول نفس الأفكار الخمس الأساسية [11]. الأفكار الكبيرة الخمس هي:
  • الإدراك: تستخدم الحواسيب المستشعرات للحصول على معلومات حول بيئتها. فهم ما تحاول الحواس إخبارنا به هو ما نسميه الإدراك.
  • التمثيل والتفكير: تحتفظ الوكلاء بنماذج للعالم وتستخدمها لاتخاذ القرارات. التمثيلات هي القوة الدافعة وراء التفكير، ويستخدمها المفكرون.
  • التعلم: يمكن للحواسيب الاستمرار في التعلم من البيانات. من خلال تعديل التمثيلات داخل شجرة القرار أو الشبكة العصبية، ينشئ خوارزمية تعلم الآلة مفكرًا.
  • التفاعل الطبيعي: للتواصل مع الناس بطريقة طبيعية، تحتاج الوكلاء الذكية إلى الوصول إلى مجموعة واسعة من المعلومات. تشمل المعلومات الفطرة السليمة، والثقافة، والعواطف البشرية، ومعرفة اللغة.
  • الأثر الاجتماعي: سيكون هناك آثار إيجابية وسلبية للذكاء الاصطناعي على المجتمع. تشمل المواضيع الآثار الاقتصادية للأتمتة، والعدالة والشفافية في أنظمة اتخاذ القرار الآلي، والاعتبارات الثقافية لخوارزميات الذكاء الاصطناعي، واستخدام الذكاء الاصطناعي من أجل الخير الاجتماعي.
تساعد هذه الأفكار الكبيرة المعلمين والباحثين في تحديد أهداف تعليم الذكاء الاصطناعي المصمم وتحديد المحتوى المطلوب. وفقًا لذلك، استخدم هذا الإطار نهج المحتوى والمنتج.
الإطار الثاني الذي تم مناقشته اقترحه تشيو وآخرون [8] في مشروعهم AI4future. قام أساتذة التعليم والهندسة بتصميم الإطار بالتعاون مع معلمي المدارس للطلاب في المرحلة المتوسطة. توضح الشكل 1 إنفوجرافيك يقدم منهج الذكاء الاصطناعي المقترح. في قلب الدائرة، يبدأ المنهج بمقدمة عن الذكاء الاصطناعي – البيانات الضخمة، وتعلم الآلة، والحوسبة السحابية. تركيز رئيسي آخر هو القضايا الأخلاقية في استخدام تطبيقات الذكاء الاصطناعي وكذلك تأثيرها الاجتماعي. الدائرة الوردية الوسطى تصور تغطيتنا لفروع مختلفة من الذكاء الاصطناعي: الذكاء الآلي الإدراكي، على سبيل المثال، “الرؤية” و”السمع”، تقنيات اللغة البشرية، على سبيل المثال، “التحدث” و”القراءة والكتابة”، الذكاءات المتكاملة، على سبيل المثال، التفكير الآلي، والمحاكاة لحل المشكلات، وإنشاء المحتوى وتوليده. الدائرة الخضراء الخارجية تصور العديد من التطبيقات المدعومة بالذكاء الاصطناعي، والتي لديها العديد من الآثار الاجتماعية الكبيرة، خاصة بالنسبة للقوى العاملة المستقبلية. وبالمثل، يركز الإطار الأول أيضًا على نهج المحتوى والمنتج، مما يساعد معلمي الذكاء الاصطناعي في إنشاء محتوى التعلم والتعليم الخاص بهم.
تم استنباط الإطار الأخير من وجهات نظر المعلمين في مشروع شراكة بين المدرسة والجامعة. أضاف تشيو [6] علم التربية إلى الإطار؛ انظر الشكل 3. يحتوي الإطار على ثلاث طبقات، موضحة باللون الأزرق الفاتح، والأبيض، والأزرق الداكن. يُظهر جوهر النموذج باللون الأزرق الفاتح ثلاثة مكونات أساسية يجب تضمينها في تعليم الذكاء الاصطناعي لطلاب المدارس المتوسطة. تتضمن مكونات المحتوى ما هو الذكاء الاصطناعي، وكيف يعالج الذكاء الاصطناعي البيانات، وما هو تأثير الذكاء الاصطناعي. تشير الطبقة الوسطى باللون الأبيض إلى ثلاث استراتيجيات تربوية: صلة الطالب، والتواصل بين المعلم والطالب، والمرونة. هذه العناصر ضرورية للتدريس الفعال لمفاهيم ومعرفة الذكاء الاصطناعي. يجب أن يشعر الطلاب بالصلة عند التعلم.
الشكل 1. إنفوجرافيك لمناهج الذكاء الاصطناعي للمرحلة الابتدائية والثانوية [4].
الذكاء الاصطناعي، كما هو موجود في حياة الطلاب. يجب على الطلاب استخدام حلول تصميم لمشاكل حقيقية. في التواصل بين المعلم والطالب، يجب على المعلمين استخدام رسومات غير مألوفة ولغة متسقة لنقل مصطلحات الذكاء الاصطناعي والخوارزميات لأنها جديدة جدًا وتجريدية للأطفال الصغار. من المرجح أن تعزز هذه اللغة التواصل بين المعلم والطالب. العنصر الأخير هو المرونة. يمكن للمعلمين استخدام محتوى وحدات وترقية المستويات الذي يوفر مسار تعلم يوجه ويقود التعلم. الطبقة الأخيرة باللون الأزرق الداكن هي نتيجة تعليم الذكاء الاصطناعي. استخدم النموذج أدوار الطلاب في مجتمع قائم على الذكاء الاصطناعي. تشمل النتائج مستخدمي الذكاء الاصطناعي، والمطورين، والباحثين، والمصممين الأخلاقيين. تجعل هذه المقاربة النتائج واضحة للطلاب. تمت إضافة نهج العملية إلى الإطار.

فجوات البحث

تعتبر هذه الأطر الثلاثة مهمة لأنها تحدد المعرفة المحتوى التي يحتاج طلاب المدارس لتعلمها وتقترح أساليب التدريس للمعلمين. ومع ذلك، تم تطويرها في مرحلة مبكرة. يجب أن يوفر التعلم بالذكاء الاصطناعي للمرحلة K-12 المساواة في التعليم، مما يقلل الفجوة الرقمية. ترتبط الكفاءة (مدى جودة أدائك) والمعرفة (ما تعرفه) ببعضها البعض، لكنها مختلفة. تتبادل الأدبيات حول تعليم الذكاء الاصطناعي مصطلحي معرفة الذكاء الاصطناعي والكفاءة. غالبًا ما نكتفي بالمعرفة بدلاً من السعي نحو الكفاءة لأنها تتطلب جهدًا أقل. يجب أن يركز تعليم الذكاء الاصطناعي على كفاءة الذكاء الاصطناعي بسبب طبيعته المدمرة. مع ظهور أدوات الذكاء الاصطناعي التوليدية مثل ChatGPT وSora، يجب إعادة النظر في معرفة وكفاءة الذكاء الاصطناعي. تعتبر العقليات التأملية الذاتية ومهارات التعلم مدى الحياة مهمة لتعليم الذكاء الاصطناعي. لذلك، من الضروري إعادة فحص هذه الأعمال الثلاثة من أجل إطار عمل أكثر شمولاً من منظور خبراء الذكاء الاصطناعي. المعلمين، حيث يمكن أن تساعد وجهات نظر المعلمين في تحسين الأطر الحالية.

هذه الدراسة والطريقة

هدف البحث

الهدفان الرئيسيان من هذه الورقة هما (1) تعريف معرفة الذكاء الاصطناعي والكفاءة لجمهور غير تقني، أي التعليم من رياض الأطفال إلى الصف الثاني عشر، و(2) اقتراح إطار عمل أكثر شمولاً لتعليم الذكاء الاصطناعي. الطريقة التي يصمم بها المعلمون أنشطة التعليم والتعلم المتعلقة بالذكاء الاصطناعي تؤثر بشكل مباشر على كيفية وما يتعلمه الطلاب. وجهات نظر المعلمين حاسمة لفهم أي تعليم مبتكر. وبناءً عليه، لتحقيق هذين الهدفين، قدمنا أولاً تعريفات معرفة الذكاء الاصطناعي والكفاءة وصممنا الإطار بالتعاون مع معلمين ذوي خبرة في الذكاء الاصطناعي. تعمل التعريفات وطرق تصميم المناهج كعدسة لإعادة فحص الأطر الثلاثة الرئيسية المتاحة في الأدبيات.

تعريفات معرفة الذكاء الاصطناعي والكفاءة

تعريف لونغ وماجيركو [31] لمحو الأمية في الذكاء الاصطناعي واسع للغاية، وربما معقد بشكل مفرط، وغير مناسب للبيئة التعليمية من رياض الأطفال حتى الصف الثاني عشر. كان من المفقود في التعريف تضمين العقليات التأملية الذاتية، التي تعتبر حاسمة للتعلم مدى الحياة حول تكنولوجيا الذكاء الاصطناعي الناشئة. من المتوقع أن تبقى تقنيات الذكاء الاصطناعي، وتتطور، وتؤثر بشكل متزايد على حياتنا، وعملنا، ودراستنا. سيتعلم المزيد من الطلاب باستخدام تقنيات الذكاء الاصطناعي المتقدمة؛ وسيعمل المزيد من الموظفين مع هذه التقنيات؛ وسيتعايش المزيد من الناس مع هذه التقنيات. للنجاح في عصر الذكاء الاصطناعي، يجب أن يكون الطلاب قادرين على تقييم فهمهم الخاص للذكاء الاصطناعي باستمرار والبقاء على اطلاع على تطوراته. وهذا يتماشى مع التوصيات التي قدمتها اليونسكو حول كفاءة الذكاء الاصطناعي للمعلمين والطلاب: “المعرفة والمهارات والمواقف التي يجب أن يكتسبها الطلاب لفهم والتفاعل بنشاط مع الذكاء الاصطناعي بطريقة آمنة وأخلاقية في المدرسة وخارجها.”.
قامت هذه الدراسة بمراجعة تعريف محو الأمية في الذكاء الاصطناعي لتحسين ملاءمته للتعليم من رياض الأطفال حتى الصف الثاني عشر، وعرفت كفاءة الذكاء الاصطناعي من خلال إضافة جوانب ثقة الطلاب وطرق التفكير الانعكاسية إلى محو الأمية في الذكاء الاصطناعي.

كما يلي:

  • يتم تعريف معرفة الذكاء الاصطناعي بأنها “قدرة الفرد على شرح كيفية عمل تقنيات الذكاء الاصطناعي وتأثيرها على المجتمع بوضوح، بالإضافة إلى استخدامها بطريقة أخلاقية ومسؤولة والتواصل والتعاون معها بفعالية في أي بيئة. وهي تركز على المعرفة (أي المعرفة والمهارات).”.
  • تُعرَّف كفاءة الذكاء الاصطناعي بأنها “ثقة الفرد وقدرته على شرح كيفية عمل تقنيات الذكاء الاصطناعي وتأثيرها على المجتمع بوضوح، بالإضافة إلى استخدامها بطريقة أخلاقية ومسؤولة والتواصل والتعاون معها بفعالية في أي بيئة. يجب أن يكون لديهم الثقة والقدرة على التفكير الذاتي حول فهمهم للذكاء الاصطناعي من أجل التعلم الإضافي. تركز على مدى استخدام الأفراد للذكاء الاصطناعي بطرق مفيدة.”
في هذه الورقة، استخدمنا التكنولوجيا، التأثير، الأخلاق، التعاون، والتأمل الذاتي لتصنيف الثقة والقدرات في تعريف كفاءة الذكاء الاصطناعي لتحليل البيانات المجمعة، كما يلي:
  • التكنولوجيا: الثقة والقدرة على شرح كيفية عمل تقنيات الذكاء الاصطناعي بوضوح
  • الأثر: الثقة والقدرة على شرح كيفية تأثير تقنيات الذكاء الاصطناعي على المجتمع بوضوح
  • الأخلاقيات: الثقة والقدرة على استخدام تقنيات الذكاء الاصطناعي بطريقة أخلاقية ومسؤولة
  • التعاون: الثقة والقدرة على التواصل والتعاون بفعالية مع تقنيات الذكاء الاصطناعي في أي بيئة
  • التأمل الذاتي: الثقة والقدرة على التأمل الذاتي في فهمهم للذكاء الاصطناعي من أجل التعلم الإضافي. الأفراد الذين يمتلكون عقولاً تأملية أقوى هم أكثر احتمالاً لمراجعة معرفتهم بالذكاء الاصطناعي وتحديد المجالات والاحتياجات للتعلم الإضافي.

المشاركون

كان هناك ما مجموعه 30 معلمًا مشاركًا في هذه الدراسة، موزعين بالتساوي بين 15 مدرسة متوسطة في هونغ كونغ (معلمين لكل مدرسة). كان متوسط عمر المعلمين 32 عامًا، وكان هناك 24 مشاركًا من الذكور و6 من الإناث. كان لدى جميعهم خبرة لا تقل عن ثلاث سنوات في تدريس الذكاء الاصطناعي. شملت خلفياتهم الأكاديمية علوم الحاسوب، الرياضيات، العلوم، الأعمال، التصميم، والتكنولوجيا. كان لكل منطقة مجموعة من المدارس، التي تفاوتت في الجودة الأكاديمية والوضع الاجتماعي. كان هناك مشارك رئيسي آخر وهو باحث يعمل كعضو هيئة تدريس ولديه خلفية أكاديمية في تعليم العلوم والتكنولوجيا والهندسة والرياضيات، وخاصة في الرياضيات والتكنولوجيا. لدى الباحث خمس سنوات من الخبرة في تعزيز تعليم الذكاء الاصطناعي في المدارس المتوسطة من خلال إنشاء مواد تعليمية، واختبار فعاليتها، وتقديم تدريب للمعلمين. بالنظر إلى أن تعليم الذكاء الاصطناعي للمرحلة K-12 قد بدأ في عام 2019، كان المشاركون ذوي خبرة.

تصميم البحث وإجراءاته

استخدمت هذه الدراسة نهج التصميم المشترك لاقتراح الإطار. كان الباحث مشاركًا في عمليات التصميم وعمل مع المعلمين. تم إجراء دورة تصميم مشترك تكرارية تضمنت جمع البيانات، والتحليل، والتصميم. ساهم الباحث والمعلمون في تطوير الإطار الشامل من خلال التعريفات، وطرق التصميم، وثلاثة أطر رئيسية موجودة تم مناقشتها في مراجعة الأدبيات.
بدأ الباحث الرئيسي بإجراء ورشة عمل مدتها 3 ساعات حول الأساليب الأربعة، والأطر الثلاثة الرئيسية، وتعريفات معرفة الذكاء الاصطناعي والكفاءة لجميع المعلمين. كانت هناك أربع دورات لجمع البيانات. في كل دورة، عقد الباحث اجتماعًا مع المعلمين وجمع تعليقاتهم. تم تقسيم المعلمين إلى خمس مجموعات من ستة وكتبوا وصححوا الإطار وتجارب التعلم من الدورة السابقة (استخدمت الدورة الأولى الأطر الثلاثة الرئيسية التي تم مناقشتها في المراجعة). استمر كل اجتماع لمدة 4 ساعات مع
استراحة لمدة 15 دقيقة. وبالتالي، اقترح الباحث والمشاركون معًا كيفية تصميم وتقديم تعليم الذكاء الاصطناعي، واستخدم المعلمون المواد كأدلة لدعم أفكارهم.

التحليلات، الموثوقية والصلاحية

في كل من الاجتماعات (أي، الدورات)، قام جميع المشاركين بتصميم الإطار وتجربة التعلم بالتعاون مع الباحث. في نهاية كل دورة، صوت المعلمون المشاركون على التغييرات في الإطار وتجارب التعلم باستخدام منصة رقمية للاستجابة الفورية. استخدمنا مستوى اتفاق بنسبة 75 في المئة (توافق)، متماشياً مع دراسات سابقة مشابهة [41]. يشير هذا المستوى إلى موثوقية عالية للتحليل.
استخدمنا ستة جوانب لمناقشة صلاحية هذا التصميم والتحليل النوعي [15،30]. الملاءمة: أولاً، استخدمنا الأطر الثلاثة الرئيسية الموجودة في الأدبيات حول تعليم الذكاء الاصطناعي. ثانياً، قام الباحث الرئيسي بتصميم وتقديم تعليم الذكاء الاصطناعي للمرحلة K-12 منذ عام 2018 ونشر أكثر من 15 ورقة بحثية تتعلق بتعليم الذكاء الاصطناعي. كان المشاركون معلمين ذوي خبرة في تعليم الذكاء الاصطناعي. هذه الأمور تظهر أن الطرق والمواد المستخدمة لجمع البيانات كانت مناسبة للسياق وسؤال البحث. التثليث: تم تجنيد معلمين من مواد ومدارس مختلفة لتجنب التحيزات الفردية (الموضوعية) وتقديم وجهة نظر أكثر تنوعاً. المصداقية: تتحدد مصداقية البحث النوعي بكفاءة الباحث في عملية البحث. كان لدى المؤلفين خبرة في البحث والتصميم وتعليم الذكاء الاصطناعي في مناطق مختلفة. الإجراءات الأخلاقية: منحت الجامعة التي ينتمي إليها المؤلفون الرئيسيون إذنًا أخلاقيًا لهذا العمل، مع معالجة الآثار الأخلاقية لجمع هذه البيانات. فهم جميع المشاركين حقوقهم وقدموا موافقتهم. القابلية للنقل: كانت المواد التعليمية الرئيسية للمشاركين متنوعة، مما يوفر نطاقًا واسعًا محتملًا من البيانات. تحقق المستجيب: تم دعم هذه الدراسة من قبل الأطر الثلاثة الرئيسية لتعليم الذكاء الاصطناعي؛ وبالتالي، فهي موثوقة وصالحة.

النتائج والمناقشات

تم استخدام الثقة والقدرات وأربعة نهج لتصميم المناهج كإطار مفاهيمي في هذه الدراسة لتحليل البيانات التي تم الحصول عليها في كل دورة. وقد حددت الاجتماعات الأولى اثنين المكونات الرئيسية الخمسة للإطار الشامل المقترح؛ انظر الشكل 3. وهي التكنولوجيا، التأثير، الأخلاق، التعاون، والتأمل الذاتي. يتم شرح كل من النتائج بالتفصيل أدناه.

المكونات الخمسة في الإطار الشامل

التكنولوجيا

المكون الأول يتعلق بالمعرفة الأساسية للذكاء الاصطناعي، والتي يمكن مقارنتها بالإطارين المقترحين من قبل الأفكار الكبيرة الخمس في الذكاء الاصطناعي وAI4future. المحتوى أمر حاسم لتصميم التعليم الشامل، أي التعليم من الروضة حتى الصف الثاني عشر. التعليم من الروضة حتى الصف الثاني عشر، على عكس التعليم العالي، يتطلب حدودًا. يجب أن تعمل المدارس المختلفة في نفس المنطقة على نفس أهداف التعلم. وفقًا للمعلمين، تشمل المعرفة الأساسية ثلاثة مواضيع: المكونات الأساسية، الإدراك، والتطبيقات.
  • يجب تدريس المكونات الأساسية، والتعريف، والتاريخ، وتطور الذكاء الاصطناعي في المدارس. تم الاتفاق على التعريف التالي للذكاء الاصطناعي من قبل جميع المشاركين: “يشير الذكاء الاصطناعي إلى قدرة الآلة على أداء مهام تعادل التعلم واتخاذ القرار البشري.” يتماشى هذا التعريف مع نتائج تشيو وآخرين وتوريتسكي وآخرين. يجب على الطلاب فهم المفاهيم الأساسية مثل البيانات الضخمة، وتعلم الآلة، والحوسبة السحابية. يجب عليهم فهم الخصائص الخمس الأساسية والفطرية للبيانات الضخمة، وهي السرعة، والحجم، والقيمة، والتنوع، والموثوقية. لفهم كيفية استخدام آلات الذكاء الاصطناعي للبيانات لتعزيز مهاراتها بشكل صحيح، يجب أن يتناول موضوع تعلم الآلة نماذج التدريب وخوارزميات التعلم. الحوسبة السحابية مطلوبة.
    لإجراء معالجة بيانات ضخمة من أجل تدريب النماذج و/أو الخوارزميات بشكل أفضل. علاوة على ذلك، فإن تاريخ وتطور الذكاء الاصطناعي هما موضوعان حاسمان في التعليم من الروضة حتى الصف الثاني عشر. اتفق جميع المشاركين على أنه يجب على الطلاب فهم تاريخ الذكاء الاصطناعي بالإضافة إلى التقدمات المعاصرة مثل “الثورة الصناعية الرابعة” والذكاء الاصطناعي التوليدي، مثل ChatGPT وSora. يجب على الطلاب أن يدركوا كيف تختلف آلات الذكاء الاصطناعي عن الآلات غير الذكية. على سبيل المثال، تم تصميم الآلات غير الذكية للإجابة على مشاكلنا من خلال تطبيق القواعد أو الخوارزميات، بينما تستخدم آلات الذكاء الاصطناعي البيانات لتطوير وتجديد القواعد أو النماذج. يجب أن يفهموا أن الذكاء الاصطناعي يغير المفهوم الأساسي لكيفية عمل الآلات وأن “البيانات هي الشيفرة الجديدة”. يجب أن يكون الطلاب الذين أتقنوا المعرفة الأساسية، على وجه الخصوص، قادرين على التعرف على ما إذا كانت التقنيات التي يستخدمونها هي ذكاء اصطناعي وفهم تداعيات ذلك. يجب أن يُطلب منهم أيضًا وصف أنواع البيانات التي يجمعها الذكاء الاصطناعي، وكيف يحلل الذكاء الاصطناعي البيانات، وكيف يتعلم الذكاء الاصطناعي من البيانات.
  • الإدراك هو الموضوع الثاني في المعرفة الأساسية، وهو متسق مع أبحاث تشيو وآخرون [8] وتوريتسكي وآخرون [42]. أشار المعلمون إلى أن مفهوم الإدراك في دراسة توريتسكي وآخرون [42] مجرد جدًا. لم يفهم الطلاب ما تعنيه هذه الكلمة. اختاروا مفاهيم الحواس البشرية كمواضيع فرعية للإدراك. الحواس البشرية – الرؤية، القراءة، الكتابة؛ التحدث والاستماع؛ التفكير والإبداع؛ الاستدلال والمحاكاة – هي مصطلحات تعكس تعريف الذكاء الاصطناعي. نظرًا لأن هذه المصطلحات ليست تقنية بشكل مفرط، سيفهم كل من الطلاب والمعلمين ما يحتاجون إلى تعلمه أو تدريسه. يجب أن يفهم الطلاب كيف تجمع كل حاسة البيانات وتعالجها.
  • الموضوع الثالث هو تطبيقات الذكاء الاصطناعي. أشار المعلمون إلى أن نطاق الموضوع أكبر من عمقه. يجب أن يكون لدى الطلاب فهم شامل لتطبيقات الذكاء الاصطناعي، والتي يجب أن تشمل معظم الصناعات أو جوانب الحياة اليومية مثل الرعاية الصحية، والترفيه، والنقل، واللوجستيات، وما إلى ذلك. يجب عليهم استخدام الإدراك لوصف كيفية عمل كل تطبيق، بالإضافة إلى التعلم الآلي لبناء وتطوير تطبيقات الذكاء الاصطناعي الخاصة بهم.

أثر

المكون الرئيسي الثاني يقترح ثلاثة مواضيع تعزز ثقة الطلاب وقدرتهم على شرح تأثير تقنيات الذكاء الاصطناعي على المجتمع بوضوح: مستقبل العمل، الخير الاجتماعي، والمخاطر.
  • في موضوع مستقبل العمل، يولي الأطفال الصغار قيمة عالية لدراستهم المستقبلية ووظائفهم. يحتاجون إلى فهم أن المزيد من المهن المستقبلية تتطلب معرفة بالذكاء الاصطناعي وأنهم من المرجح بشكل متزايد أن يتعلموا مع الذكاء الاصطناعي في حياتهم ويعملوا مع الذكاء الاصطناعي في وظائفهم.
  • الموضوع الفرعي الثاني هو الذكاء الاصطناعي من أجل الخير الاجتماعي. يمكن أن يكون للذكاء الاصطناعي آثار إيجابية وسلبية على المجتمع. وفقًا للمعلمين، يجب على الطلاب أن يتعلموا كيف يحل الذكاء الاصطناعي القضايا المعقدة من خلال معالجة القضايا الاجتماعية والبيئية والصحية العامة الحرجة. بخلاف الفوائد التي يجلبها الذكاء الاصطناعي للمجتمع، يجب على الطلاب أيضًا أن يتعلموا المخاطر.
  • الموضوع الأخير هو المخاطر. اتفق جميع المشاركين على أنه يجب على الطلاب أن يكونوا على دراية بالمخاطر المحتملة المرتبطة بالذكاء الاصطناعي. يجب عليهم التحقيق في كيفية تسبب التقنيات الناشئة في المشاكل والأذى في سياقات مختلفة. هذه تتماشى مع الإطارين [4،43] ودراسات أخرى ذات صلة [47،50].

الأخلاق

المكون الثالث هو الأخلاق، وهناك مخاوف بشأن أخلاقيات الذكاء الاصطناعي والتحيز البشري. ناقش المعلمون أي المفاهيم يجب تضمينها في التعليم من الروضة حتى الصف الثاني عشر. المبادئ أو السياسات الحالية المنشورة للجمهور معقدة للغاية. تم استخدام بعضها لأغراض قانونية وتجارية. وأعربوا عن أن “سياسات الذكاء الاصطناعي تشمل 20 مجالًا للاستخدام الأخلاقي.” إنها معقدة جدًا للأطفال الصغار أو غيرهم.
الجماهير التقنية. استخدم المعلمون أخيرًا مبادئ الذكاء الاصطناعي الأخلاقية من IBM كنقطة انطلاق وتوصلوا إلى توافق حول ما يجب تضمينه في هذا الموضوع. اختاروا خمسة منها لأنها أكثر صلة بطلاب المدارس: العدالة والتحيزات؛ الثقة والشفافية؛ المساءلة؛ الفائدة الاجتماعية؛ والخصوصية والأمان.

التعاون

المكون الرابع هو تعزيز ثقة الطلاب وقدرتهم على التواصل والتعاون مع تقنيات الذكاء الاصطناعي في أي بيئة. أشار المعلمون المشاركون إلى أن هذه الثقة والقدرة كانت مرتبطة بالثلاثة مواضيع التي تم مناقشتها سابقًا: التكنولوجيا، التأثير، والأخلاقيات. اقترحوا أن “يحتاج الطلاب إلى تعلم كيفية طرح الأسئلة بفعالية عند استخدام ChatGPT”، و”يجب على الطلاب فحص موثوقية توقعات نظام التنبؤ بالذكاء الاصطناعي”، و”يجب أن يكون الطلاب على دراية بحقوقهم في الخصوصية عند استخدام أنظمة الذكاء الاصطناعي”، و”يجب أن يكون الطلاب على علم بأن أنظمة الذكاء الاصطناعي قد تجمع بياناتهم، مثل المواقع والأصوات.” من المحتمل أن يتم تعزيز ذلك من خلال دراسات الحالة، والتعلم القائم على المشاريع، والأنشطة العملية. بشكل عام، يتماشى هذا مع بعض الدراسات حول هندسة الطلب، وهي التقنية التي تهدف إلى هيكلة النص بحيث يمكن لنموذج الذكاء الاصطناعي التوليدي فهمه واستيعابه. المروج هو نص بلغة طبيعية يمكّن من الحصول على استجابات أكثر ملاءمة من النموذج.

التأمل الذاتي

المكون الرئيسي الأخير يتعلق بعقليات الطلاب الذاتية الانعكاسية التي ترتبط ارتباطًا وثيقًا بالثقة. الطالب الواثق يكون أكثر ميلًا للتفكير الذاتي باستمرار حول معرفته بالذكاء الاصطناعي مقارنة بالمتعلمين الأقل ثقة. نظرًا لأن الذكاء الاصطناعي هو تقنية ناشئة وم disruptive، فمن الضروري أن يقوم الطلاب بتقييم معرفتهم بالذكاء الاصطناعي باستمرار للبقاء على اطلاع. هذا يتعلق بالمعرفة ولكنه متأثر بجوانبه العاطفية، حيث إن الانخراط المعرفي مرتبط بالانخراط العاطفي. على سبيل المثال، الطلاب الذين يشعرون بحماس أكبر تجاه الذكاء الاصطناعي أو التكنولوجيا هم أكثر احتمالًا لقراءة المقالات ومشاهدة الفيديوهات حول أحدث التطورات في التكنولوجيا المتعلقة بالذكاء الاصطناعي. هذا مهم جدًا لمهارات التعلم مدى الحياة في مجتمع يعتمد على الذكاء الاصطناعي.
ركزت الاجتماعات الأخيرة في الدراسة على مناقشة العمليات وطرق التصميم القائمة على الممارسة. توصل المعلمون إلى توافق في الآراء بأن هناك خمس تجارب تعليمية أساسية يجب أخذها في الاعتبار عند تصميم أنشطة التعليم والتعلم باستخدام الذكاء الاصطناعي.

خمسة تجارب تعليمية أساسية

تم تحديد خمس تجارب تعليمية لتعليم الذكاء الاصطناعي في الاجتماعات. اتفق جميع المعلمين على أن الخبرة ضرورية لتعلم الذكاء الاصطناعي. يناقش ما يلي كيف تعزز التجارب التعليمية التعليم الشامل والمتنوع في مجال الذكاء الاصطناعي.
  • المشاركة المجتمعية: وفقًا لكوبر [13] وموني وإدواردز [36]، فإن دمج تعلم الطلاب مع المجتمع هو استراتيجية بيداغوجية هادفة يستخدمها المعلمون لإقامة صلة بين ما يتم تدريسه في الفصل الدراسي ومجتمعات الطلاب المحلية. ونتيجة لذلك، يمكن للطلاب تطبيق أفكارهم بناءً على الملاحظة الشخصية والتفاعل الاجتماعي لتصميم وإيجاد حلول لمشاكل العالم الحقيقي في المجتمع. وهم أكثر ميلاً للاستثمار في التعلم لأن التحديات أكثر صلة وتشجع على مشاركة الطلاب. ستزيد هذه المشاركة المجتمعية من اهتمام الطلاب وحماسهم. ، 28]، مما يؤدي إلى تعليم الذكاء الاصطناعي الأكثر شمولاً وتنوعاً [46]. ستعزز هذه المشاركة المجتمعية الذكاء الاصطناعي من أجل الخير الاجتماعي بينما تنمي أيضاً المواقف الإيجابية للطلاب تجاه الذكاء الاصطناعي. علاوة على ذلك، ذكر المعلمون أن نهج التفكير التصميمي – التعاطف، التعريف، التفكير الإبداعي، النموذج الأولي، والاختبار – يجب أن يُستخدم لحل مشاكل المجتمع. لتحقيق تأثير أكبر، يمكن أن تتعلم المشاركة المجتمعية الكثير من التفكير التصميمي [4].
  • دراسات حالة عالمية ومحلية: واحدة من الطرق التعليمية الفعالة التي اقترحها المعلمون هي استخدام المقالات الإعلامية لتثقيف الطلاب حول أخلاقيات الذكاء الاصطناعي وتأثيره. يمكن لطلاب المدارس قراءة مقالات ويب متنوعة، ومع ذلك هناك العديد من الخيارات المتحيزة أو الأخبار الكاذبة. هم أقل نضجًا وأقل قدرة على الحكم على موثوقية المقالات. على عكس ما كان عليه الحال قبل عشر سنوات، يتم نشر المزيد من القصص الصحفية نتيجة لتقدم الذكاء الاصطناعي. لقد أصبحت مؤخرًا أكثر سهولة لكل من المعلمين والطلاب. هذه الطريقة تخلق تعلمًا غير تقليدي، مما يجعلها أكثر أصالة وملاءمة. يمكن للطلاب أن يعكسوا بشكل أكثر نشاطًا على القضايا الأخلاقية التي تثيرها القصص الصحفية. علاوة على ذلك، يمكن استخدام هذا الأسلوب التعليمي لتعليم المبادئ الأخلاقية مع التأكيد على أهمية مصادر البيانات. يمكن للمعلمين استخدام ذلك لتوضيح مفاهيم مثل العدالة والتحيز، والثقة والشفافية، والمساءلة.
  • الأنشطة العملية: الطلاب محاطون بالذكاء الاصطناعي، لكن ليس كثيرًا في فصولهم الدراسية. يمكن أن تساعد الأنشطة العملية الطلاب على فهم الإدراك بشكل أفضل لأنها تضعهم في مقعد السائق من خلال النشاط البدني والتعلم النشط. يمكن أن يدعم التعلم النشط بدنيًا ثقة الطلاب وقدرتهم على نمذجة العالم وتوليد أفكار إبداعية. قد تسمح هذه الروابط بتصميم واستخدام أدوات متعددة الأجزاء. يمكن للطلاب حل المشكلات التي لا يمكنهم حلها شفهيًا أو بصريًا من خلال نمذجة الأنظمة الفيزيائية بأيديهم. إن طلب من الطلاب استخدام أيديهم لنمذجة الظواهر الفيزيائية باستخدام الذكاء الاصطناعي سيطور فهمهم المفاهيمي لإدراك الذكاء الاصطناعي. كما اقترح المعلمون بعض الأفكار العملية: يمكن للطلاب التحقيق في كيفية جمع الذكاء الاصطناعي وفهمه وتحديده للصور، أي تعلم كيفية رؤية الحواسيب للصور. يمكن للطلاب فهم ما يقرؤون ويكتبون نظرًا لأن الذكاء الاصطناعي يفهم اللغات والنصوص، ويمكنهم تطوير محللات نصية لاكتشاف مزاج النصوص. علاوة على ذلك، يعتقد معظم الطلاب أن الذكاء الاصطناعي والروبوتات هما نفس التكنولوجيا. يجب على الطلاب أن يتعلموا كيف يختلف الذكاء الاصطناعي عن الروبوتات من خلال القيام بمشروع عملي يبنون فيه روبوتًا مع التفكير والإدراك. وفقًا للمعلمين، يمكن تعليم المبادئ الأخلاقية، التي غالبًا ما تم تدريسها باستخدام أساليب دراسة الحالة، من خلال الأنشطة العملية. يمكن للطلاب تصميم تطبيقات ذكاء اصطناعي متحيزة واحدة أو أكثر وشرح كيف أدى اختيار مجموعة البيانات إلى النتائج المتحيزة. نتيجة لذلك، يمكن أن تلبي الأنشطة العملية بشكل أفضل احتياجات الطلاب ذوي الفروق الفردية والتعلم وتعزز التعليم الشامل والمتنوع في مجال الذكاء الاصطناعي.
  • المعارض والعروض التقديمية: هذه الاستراتيجية شائعة في نشر أعمال الطلاب، لا سيما في التعلم القائم على المشاريع. إنها تتيح للطلاب توطيد تعلمهم من خلال التواصل حول عملياتهم، والتفكير في منتجاتهم، والتأمل في إجاباتهم. أشار المعلمون في هذا الإطار إلى أنه يجب على الطلاب عرض عملية تعلمهم وأعمال مشاريعهم. وأبرز المعلم في تعليم الذكاء الاصطناعي أنه يجب على الطلاب استخدام المعرفة الأساسية كمعايير لتقديم أعمالهم في المعرض. على سبيل المثال، كيف حصلوا على البيانات أو أنشأوها لبناء الإدراك؟ هل البيانات أخلاقية؟ هل استخدامهم لها أخلاقي؟ ما هو الأثر الاجتماعي لعملهم؟ أي صناعة ستستفيد من منتجاتهم؟ وبناءً عليه، فإن تقديم أعمال مشاريعهم يشجع الطلاب على الاستفادة من اختلافاتهم الفردية.
  • التعلم الثقافي: يجب أن تؤخذ القيم الإنسانية والثقافة في الاعتبار عند تعلم الذكاء الاصطناعي. قد تكون البيانات متحيزة بسبب كيفية الحصول عليها أو اختيارها للاستخدام. تلعب القيم الإنسانية والثقافة دورًا كبيرًا في ذلك. قد لا يتم تحديد الإجابة على سؤال “هل هناك تحيز في الذكاء الاصطناعي أم لا” من خلال معرفة الذكاء الاصطناعي ولكن من خلال القيم الإنسانية والثقافة. يجب أن تؤخذ هذان العنصران في الاعتبار من قبل الطلاب أثناء تعلم الذكاء الاصطناعي. على سبيل المثال، عند تطوير مشاريع الذكاء الاصطناعي لمساعدة كبار السن، يتعين على الطلاب احترام ثقافتهم وقيمهم (مثل الطعام والأنظمة الغذائية) ولكن لا يغيرون ثقافتهم. عند مناقشة القضايا الأخلاقية المتعلقة بالسيارات ذاتية القيادة، يجب أن يكون الطلاب على دراية بالقوانين المحلية والمعتقدات الدينية (إيذاء الأبقار في الهند قد يؤدي إلى جرائم خطيرة).
بشكل عام، ستعزز توصيات المعلمين لخمس تجارب تعلم أساسية التعليم الشامل والمتنوع في الذكاء الاصطناعي من خلال
الخير الاجتماعي (مشاركة المجتمع)، صلة الطلاب (مقالات الصحف)، التعلم النشط (أنشطة عملية)، التعاون، والتواصل (عرض تقديمي) (الشكل 2).

اتجاهات البحث المستقبلية والتوصيات

بالمقارنة مع الأطر الرئيسية الحالية لتعليم الذكاء الاصطناعي في المرحلة الابتدائية والثانوية، يحتوي هذا الإطار على جانبين جديدين مضافين – الثقة والعقلية الانعكاسية الذاتية – وموهبتين جديدتين: (1) التعاون: القدرة على التواصل والتعاون بفعالية مع تقنيات الذكاء الاصطناعي في أي بيئة؛ (2) الانعكاس الذاتي: القدرة على الانعكاس الذاتي لفهمهم للذكاء الاصطناعي لمزيد من التعلم. يركز هذا البحث على مناقشة خمسة اتجاهات بحث مستقبلية لمعالجة هذه الثقة والقدرات.
  • هندسة الطلب: في هذه التقنية الجديدة، يمكنك إعادة صياغة استفسار، اختيار أسلوب، تقديم مزيد من المعلومات السياقية، أو تعيين دور للذكاء الاصطناعي عند التفاعل مع . تقتصر الدراسات الحالية حول هذه التقنية على ChatGPT. يتعلق هذا أيضًا بمهاراتنا في طرح الأسئلة. يتم تطوير معظم مهارات طرح الأسئلة لدى الأطفال الصغار في التعليم في المرحلة الابتدائية والثانوية. يجب إجراء مزيد من الدراسات لفهم كيفية تفاعل طلاب المرحلة الابتدائية والثانوية مع الذكاء الاصطناعي، أي، هندسة الطلب أو مهارات طرح الأسئلة للذكاء الاصطناعي.
  • تعليم علم البيانات: كيفية التواصل مع الذكاء الاصطناعي مرتبط بمن يستخدم البيانات ويفسرها. ترتبط محو الأمية البيانات ارتباطًا وثيقًا بمحو الأمية في الذكاء الاصطناعي [31]. قد يحتاج تعليم علم البيانات ومحو الأمية البيانات إلى أن يتم تضمينها في التعليم في المرحلة الابتدائية والثانوية. حاليًا، يتم تضمين هذه الأمية في مواد الرياضيات أو الذكاء الاصطناعي. يجب إجراء مزيد من الدراسات للتحقيق في كيفية ارتباط محو الأمية البيانات بتطوير محو الأمية في الذكاء الاصطناعي.
  • محو الأمية الخوارزمية: للتواصل الكامل مع الذكاء الاصطناعي، قد تكون محو الأمية الخوارزمية مطلوبة. تشمل هذه الأمية الوعي والمعرفة بالخوارزميات، والثقة والاطمئنان للخوارزميات، وتقدير الخوارزميات، وتجنب الخوارزميات [25]. لم يناقش هذا الإطار هذه الأمية. يجب إجراء دراسات مستقبلية لتعريف محو الأمية الخوارزمية وكيف تؤثر على محو الأمية في الذكاء الاصطناعي.
  • العقليات الانعكاسية الذاتية: هذه العقلية لم يتم دراستها بشكل كافٍ في تعليم الذكاء الاصطناعي. الذكاء الاصطناعي مختلف عن المواد الأخرى التقليدية، مثل الرياضيات والعلوم في المرحلة الابتدائية والثانوية، وهو في طور النشوء. ترتبط النظرية الاجتماعية المعرفية بالانخراط العاطفي، والتعلم الذاتي المنظم يتعلق بالانعكاس الذاتي. يُوصى بإجراء دراسات مستقبلية لاستخدام النظرية الاجتماعية المعرفية والتعلم الذاتي المنظم لتحديد أساليب التدريس الفعالة لتعزيز العقليات الانعكاسية الذاتية لدى الطلاب.
  • البحث التجريبي: لا يزال تعليم الذكاء الاصطناعي جديدًا؛ يجب إجراء مزيد من الدراسات التجريبية لتنقيح تعريف محو الأمية في الذكاء الاصطناعي والكفاءة وإطار تعليم الذكاء الاصطناعي.

الخاتمة والقيود

يبدو أن تصميم تعليم فعال في الذكاء الاصطناعي في المرحلة الابتدائية والثانوية أمر صعب للغاية. تم تقديم بعض الأطر – “خمسة أفكار كبيرة في الذكاء الاصطناعي”، AI4future، ونهج شامل – في الأدبيات لمساعدة المعلمين في إنشاء مناهجهم. ومع ذلك، تركز على التعرف على محتوى الذكاء الاصطناعي، والأدوات، والقدرات التقنية، والتفكير. قد لا توفر لهم فهمًا أساسيًا بشكل صحيح وتجعلهم مستعدين لمستقبل مع الذكاء الاصطناعي. لمعالجة هذه المشكلة، عرّف هذه الدراسة كفاءة الذكاء الاصطناعي من خلال إضافة جوانب الثقة والعقليات الانعكاسية الذاتية إلى محو الأمية في الذكاء الاصطناعي واقترحت إطارًا شاملاً لتعليم الذكاء الاصطناعي في . نعتقد أن الإطار يعزز التعليم الشامل والمتنوع، مما يضمن نجاح كل طالب [39]. يمكن للمعلمين تصميم تعليم الذكاء الاصطناعي لتلبية احتياجات الطلاب [37].
نقر بأن هناك قيودين في هذه الدراسة. أولاً، لم يتم اختبار الإطار في الميدان، لذا هناك حاجة إلى مزيد من البحث للتحقيق في فائدته. إحدى الطرق للقيام بذلك هي أن تستخدم مجموعة متنوعة من المدارس (بما في ذلك تلك التي تخدم الطلاب ذوي الاحتياجات الخاصة) الإطار كدليل لتطوير وتنفيذ مبادرات تعليم الذكاء الاصطناعي. ثانيًا، لم تتناول هذه الدراسة كيف تؤثر قدرة المعلم على تصميم التعلم وتطويره، وهو قيد رئيسي نظرًا لأن معظم المعلمين لا يدرسون مواضيع الذكاء الاصطناعي في تعليمهم ولا يخضعون لتدريب رسمي في الذكاء الاصطناعي. يعد تعلم كيفية تصميم وتنفيذ برامج مهنية للمعلمين في الذكاء الاصطناعي وتأثيرها على تصميم التعلم مجالًا مهمًا للبحث المستقبلي.

بيان مساهمة المؤلفين

توماس ك.ف. تشيو: الكتابة – مراجعة وتحرير، الكتابة – المسودة الأصلية، التحقق، المنهجية، التحقيق، الحصول على التمويل، التحليل الرسمي، التصور. زبير أحمد: الكتابة – مراجعة وتحرير. مورويد إسماعيلوف: الكتابة – مراجعة وتحرير. إسماعيل تميتايو سانوسي: الكتابة – مراجعة وتحرير.

إعلان عن تضارب المصالح

يعلن المؤلفون أنهم ليس لديهم أي مصالح مالية متنافسة معروفة أو علاقات شخصية قد تكون قد أثرت على العمل المبلغ عنه في هذه الورقة.
الشكل 2. إطار لتصميم نهج شامل لتعليم الذكاء الاصطناعي في المرحلة الابتدائية والثانوية (معتمد من [5]).
الشكل 3. المكونات الرئيسية الخمسة في الإطار الشامل.

توفر البيانات

تتوفر مجموعات البيانات المستخدمة في الدراسة الحالية من المؤلف المقابل عند الطلب المعقول.

التمويل

هذا مدعوم من صندوق البحث العام (رمز المشروع: 14610522)، مجلس منح البحث، هونغ كونغ، منطقة إدارية خاصة.

الشكر

يشكر المؤلفون المعلمين في المدارس على مشاركتهم في هذه الدراسة.

References

[1] Barton AC, Tan E, Greenberg D. The makerspace movement: Sites of possibilities for equitable opportunities to engage underrepresented youth in STEM. Teach Coll Rec 2017;119(6):1-44. https://doi.org/10.1177/016146811711900608.
[2] Chiu TKF. Using Self-determination Theory (SDT) to explain student STEM interest and identity development. Instr Sci 2024;58:89-107. https://doi.org/10.1007/ s11251-023-09642-8.
[3] Chiu T.K.F. (2023). The impact of Generative AI (GenAI) on practices, policies and research direction in education: A case of ChatGPT and Midjourney, Interactive Learning Environments, Advanced online publication. https://dx.doi.org/10.1080/ 10494820.2023.2253861.
[4] Chiu TKF. Applying the Self-determination Theory (SDT) to explain student engagement in online learning during the COVID-19 pandemic. J Res Technol Edu 2022;54(1):14-30. https://doi.org/10.1080/15391523.2021.1891998.
[5] Chiu TKF. A holistic approach to Artificial Intelligence (AI) curriculum for K-12 schools. TechTrends 2021;65:796-807. https://doi.org/10.1007/s11528-021-00637-1.
[6] Chiu TKF, Chai CS, Williams J, Lin TJ. Teacher professional development on Selfdetermination Theory-based design thinking in STEM education. Edu Technol Soc 2021;24(4):153-65. https://www.jstor.org/stable/48629252.
[7] Chiu TKF, Ismailov M, Zhou X-Y, Xia Q, Au D, Chai CS. Using Self-Determination Theory to explain how community-based learning fosters student interest and identity in integrated STEM education. Int J Sci Math Educ 2023;21:109-30. https://doi.org/10.1007/s10763-023-10382-x.
[8] Chiu TKF, Meng H, Chai CS, King I, Wong S, Yeung Y. Creation and evaluation of a pre-tertiary artificial intelligence (AI) curriculum. IEEE Transac Edu 2022;65(1): 30-9. https://doi.org/10.1109/TE.2021.3085878.
[9] Chiu, T.K.F., Moorhouse, B.L., Chai, C.S., & Ismailov M. (2023). Teacher support and student motivation to learn with artificial intelligence (AI) chatbot, Interactive Learning Environments, Advanced online publication. https://doi.org/10.1080/10 494820.2023.2172044.
[10] Chiu TKF, Xia Q, Zhou X-Y, Chai CS, Cheng M-T. Systematic literature review on opportunities, challenges, and future research recommendations of artificial intelligence in education. Comput Edu: AI 2023;4:100118. https://doi.org/ 10.1016/j.caeai.2022.100118.
[11] CSTA. (2017). Computer Science Teachers Association (CSTA) K-12 Computer Science Standards, Revised 2017. https://www.csteachers.org/page/standards. Accessed 12 Nov 2022.
[12] Cooper G. Examining science education in ChatGPT: An exploratory study of generative artificial intelligence. J Sci Educ Technol 2023;32:444-52. https://doi. org/10.1007/s10956-023-10039-y.
[13] Cooper JE. Strengthening the case for community-based learning in teacher education. J Teach Educ 2007;58(3):245-55. https://doi.org/10.1177/ 0022487107299979.
[14] Cope B, Kalantzis M, Searsmith D, Cope B, Kalantzis M, Searsmith D. Artificial intelligence for education: Knowledge and its assessment in AI-enabled learning ecologies. Educational Philosophy and Theory 2020;53(12):1229-45. https://doi. org/10.1080/00131857.2020.1728732.
[15] Cypress BS. Rigor or reliability and validity in qualitative research: Perspectives, strategies, reconceptualization, and recommendations. Dimensions of Critical Care Nursing 2017;36(4):253-63. https://doi.org/10.1097/DCC.0000000000000253.
[16] Datta P, Whitmore M, Nwankpa JK. A perfect storm: social media news, psychological biases, and AI. Digital Threats: Research and Practice 2021;2(2): 1-21. https://doi.org/10.1145/3428157.
[17] Falloon G. From digital literacy to digital competence: the teacher digital competency (TDC) framework. Educational Technology Research and Development 2020;68:2449-72. https://doi.org/10.1007/s11423-020-09767-4.
[18] Gerke S, Minssen T, Cohen G. Ethical and legal challenges of artificial intelligencedriven healthcare. Artificial intelligence in healthcare. Academic Press; 2020. p. 295-336. https://doi.org/10.1016/B978-0-12-818438-7.00012-5.
[19] Ghasemaghaei M. Understanding the impact of big data on firm performance: The necessity of conceptually differentiating among big data characteristics. Int J Inf Manage 2021;57:102055. https://doi.org/10.1016/j.ijinfomgt.2019.102055.
[20] Glatthorn AA, Boschee F, Whitehead BM, Boschee BF. Curriculum leadership: strategies for development and implementation. London: SAGE; 2018.
[21] Grant MM, Branch RM. Project-based learning in a middle school: Tracing abilities through the artifacts of learning. J Res Tech Edu 2005;38(1):65-98. https://doi. org/10.1080/15391523.2005.10782450.
[22] Holbrook J, Rannikmae M. The meaning of scientific literacy. Int J Environ Sci Edu 2009;4(3):275-88.
[23] Holzmeyer C. Beyond ‘AI for Social Good'(AI4SG): social transformations-Not tech-fixes-For health equity. Interdisciplinary Science Reviews 2021;46(1-2): 94-125. https://doi.org/10.1080/03080188.2020.1840221.
[24] Hornberger M, Bewersdorff A, Nerdel C. What do university students know about Artificial Intelligence? Development and validation of an AI literacy test. Computers and Education: Artificial Intelligence 2023;5:100165. https://doi.org/ 10.1016/j.caeai.2023.100165.
[25] Du YR. Personalization, Echo Chambers, News Literacy, and Algorithmic Literacy: A Qualitative Study of AI-Powered News App Users. J Broadcast Electron Media 2023;67(3):246-73. https://doi.org/10.1080/08838151.2023.2182787.
[26] Keefe EB, Copeland SR. What is literacy? The power of a definition. Research and Practice for Persons with Severe Disabilities 2011;36(3-4):92-9. https://doi.org/ 10.2511/027494811800824507.
[27] Kelly AV. The curriculum: theory and practice. 6th ed. London: Sage; 2009.
[28] King NS, Pringle RM. Black girls speak STEM: Counterstories of informal and formal learning experiences. J Res Sci Teach 2019;56(5):539-69. https://doi.org/ 10.1002/tea.21513.
[29] Laupichler MC, Aster A, Raupach T. Delphi study for the development and preliminary validation of an item set for the assessment of non-experts’ AI literacy. Computers and Education: Artificial Intelligence 2023;4:100126. https://doi.org/ 10.1075/idj.23.1.03dighttps://doi.org/10.1016/j.caeai.2023.100126.
[30] Leung L. Validity, reliability, and generalizability in qualitative research. J Family Med Prim Care 2015;4(3):324. https://doi.org/10.4103/2249-4863.161306.
[31] Long D, Magerko B. What is AI literacy? Competencies and design considerations. In: Proceedings of the 2020 CHI conference on human factors in computing systems; 2020. p. 1-16. https://doi.org/10.1145/3313831.3376727.
[32] Liu P, Yuan W, Fu J, Jiang Z, Hayashi H, Neubig G. Pre-train, prompt, and predict: A systematic survey of prompting methods in natural language processing. ACM Comput Surv 2023;55(9):1-35.
[33] Martin A, Grudziecki J. DigEuLit: Concepts and tools for digital literacy development. Innovation in Teaching and Learning in Information and Computer Sciences 2006;5(4):249-67. https://doi.org/10.11120/ital.2006.05040249.
[34] Marsh CJ, Willis G. Curriculum: alternative approaches, ongoing issues. Upper Saddle River, NJ: Merrill/Prentice Hall; 2003.
[35] Meskó B. Prompt engineering as an important emerging skill for medical professionals: tutorial. J Med Internet Res 2023;25:e50638. https://doi.org/ 10.2196/50638.
[36] Mooney LA, Edwards B. Experiential learning in sociology: Service learning and other community-based learning initiatives. Teach Sociol 2001;29(2):181-94. https://doi.org/10.2307/1318716.
[37] Priestley M. Whatever happened to curriculum theory? Critical realism and curriculum change. Pedagogy, Culture & Society 2011;19(2):221-37. https://doi. org/10.1080/14681366.2011.582258.
[38] Priestley M, Biesta G, editors. Reinventing the curriculum: new trends in curriculum policy and practice. A&C Black; 2013.
[39] Simmons J, MacLean J. Physical education teachers’ perceptions of factors that inhibit and facilitate the enactment of curriculum change in a high-stakes exam climate. Sport Educ Soc 2018;23(2):186-202. https://doi.org/10.1080/ 13573322.2016.1155444.
[40] Tal T. Pre-service teachers’ reflections on awareness and knowledge following active learning in environmental education. International Research in Geographical and Environmental Education 2010;19(4):263-76. https://doi.org/ 10.1080/10382046.2010.519146.
[41] Teixeira PJ, Marques MM, Silva MN, Brunet J, Duda J, Haerens L, La Guardia J, Lindwall M, Lonsdale C, Markland D. A Classification of Motivation and Behavior Change Techniques Used in Self-Determination Theory-Based Interventions in Health Contexts. Motiv Sci 2020;6(4):438-55. https://doi.org/10.1037/ mot0000172.
[42] Touretzky DS, Gardner-McCune C, Martin F, Seehorn D. Envisioning AI for K-12: What should every child know about AI?. In: Proceedings of AAAI-19; 2019. https://doi.org/10.1609/aaai.v33i01.33019795.
[43] Touretzky D, Gardner-McCune C, Seehorn D. Machine learning and the five big ideas in AI. Int J Artif Intell Educ 2023;33(2):233-66. https://doi.org/10.1007/ s40593-022-00314-1.
[44] Tsarava K, Moeller K, Román-González M, Golle J, Leifheit L, Butz MV, Ninaus M. A cognitive definition of computational thinking in primary education. Comput Educ 2022;179:104425. https://doi.org/10.1016/j.compedu.2021.104425.
[45] Wolff A, Gooch D, Montaner JJC, Rashid U, Kortuem G. Creating an understanding of data literacy for a data-driven society. J Commun Inform 2016;12(3). https:// doi.org/10.15353/joci.v12i3.3275.
[46] Xia Q, Chiu TKF, Lee M, Temitayo I, Dai Y, Chai CS. A self-determination theory design approach for inclusive and diverse artificial intelligence (AI) K-12 education. Comput Educ 2022;189:104582. https://doi.org/10.1016/j. compedu.2022.104582.
[47] Williams R, Ali S, Devasia N, DiPaola D, Hong J, Kaputsos SP, Breazeal C. AI+ ethics curricula for middle school youth: Lessons learned from three project-based curricula. Int J Artif Intell Educ 2023;33(2):325-83. https://doi.org/10.1007/ s40593-022-00298-y.
[48] Yannier N, Hudson SE, Koedinger KR, Hirsh-Pasek K, Golinkoff RM, Munakata Y, Brownell SE. Active learning:”Hands-on” meets “minds-on. Science (1979) 2021; 374(6563):26-30. https://doi.org/10.1126/science.abj9957.
[49] Zhang C, Lu Y. Study on artificial intelligence: The state of the art and future prospects. J Ind Inf Integr 2021;23:100224. https://doi.org/10.1016/j. jii.2021.100224.
[50] Zhang H, Lee I, Ali S, DiPaola D, Cheng Y, Breazeal C. Integrating ethics and career futures with technical learning to promote AI literacy for middle school students: An exploratory study. Int J Artif Intell Educ 2023;33(2):290-324. https://doi.org/ 10.1007/s40593-022-00293-3.

    • Corresponding author at: Faculty of Education, The Chinese University of Hong Kong, Shatin, Hong Kong.
    E-mail addresses: tchiu@cuhk.edu.hk (T.K.F. Chiu), zubairtarar@qu.edu.qa (Z. Ahmad), ismailov.murod.gm@u.tsukuba.ac.jp (M. Ismailov), ismaila.sanusi@uef.fi (I.T. Sanusi).

Journal: Computers and Education Open, Volume: 6
DOI: https://doi.org/10.1016/j.caeo.2024.100171
Publication Date: 2024-03-13

What are artificial intelligence literacy and competency? A comprehensive framework to support them

Thomas K.F. Chiu , a , Zubair Ahmad , Murod Ismailov , Ismaila Temitayo Sanusi Department of Curriculum and Instruction, Centre for Learning Sciences and Technologies and Centre for University and School Partnership at The Chinese University of Hong Kong Module Development and Publication, Qatar University Institute of Humanities and Social Sciences, The University of Tsukuba School of Computing, University of Eastern Finland Qatar University Young Scientists Center (QUYSC), 2713, Qatar University, Doha Qatar

ARTICLE INFO

Keywords:

AI literacy
AI competency
K-12 education
Machine learning
Data literacy
Generative AI

Abstract

Artificial intelligence (AI) education in K-12 schools is a global initiative, yet planning and executing AI education is challenging. The major frameworks are focused on identifying content and technical knowledge (AI literacy). Most of the current definitions of AI literacy for a non-technical audience are developed from an engineering perspective and may not be appropriate for education. Teacher perspectives are essential to making sense of this initiative. Literacy is about knowing (knowledge, what skills); competency is about applying the knowledge in a beneficial way (confidence, how well). They are strongly related. This study goes beyond knowledge (AI literacy), and its two main goals are to (i) define AI literacy and competency by adding the aspects of confidence and self-reflective mindsets, and (ii) propose a more comprehensive framework for K-12 AI education. These definitions are needed for this emerging and disruptive technology (e.g., ChatGPT and Sora, generative AI). We used the definitions and the basic curriculum design approaches as the analytical framework and teacher perspectives. Participants included 30 experienced AI teachers from 15 middle schools. We employed an iterative co-design cycle to discuss and revise the framework throughout four cycles. The definition of AI competency has five abilities that take confidence into account, and the proposed framework comprises five key components: technology, impact, ethics, collaboration, and self-reflection. We also identify five effective learning experiences to foster abilities and confidences, and suggest five future research directions: prompt engineering, data literacy, algorithmic literacy, self-reflective mindset, and empirical research.

Introduction

Artificial intelligence (AI) is the ability of a digital machine to carry out tasks that are typically performed by intelligent beings. Its technologies that support AI includes computer vision, speech-to-text, and natural language processing [7]. The advancements in AI are having profound effects on our daily lives, entertainment, education, and jobs. It is critical to expand AI training beyond higher education and professionals. With the goal of preparing all citizens for AI based society, AI education has been included in non-expert community around the world, e.g., AI for all [24] and AI for K-12 [4,43]. We need all our young children to have good AI literacy and competency [3]. Therefore, AI education for is a global initiative, as evidenced in UNESCO’s
report on AI education. On the other hand, unlike in higher education, designing education must take into account implementation and variety in delivery. To address the global initiative, a few major frameworks were proposed for researchers and educators [4,42,43]. They focus on identifying content and knowledge for AI teaching and learning. They are very important at the beginning of this initiative because researchers and educators did not know what to include in the AI curriculum. However, AI education is more than content, because it addresses the learning outcomes (i.e., What are AI literacy and competency?) and experiences (i.e. How to nurture them?) [5,27,34].
Literature has defined and suggested what AI literacy is for non-AI professionals (e.g., [29,31]). Younger children might not benefit from the definition that was developed via an engineering perceptive. Long
and Margerko [31] stated that their definition needs to be reviewed in the future as research on AI education is still in its early stages. The technologies are emerging and disruptive, such as ChatGPT and Sora. These generative AI tools keep humans learning [9]. Therefore, this study argues that (i) the concepts of self-reflective mindsets should be included in the definition of AI literacy (i.e., AI competency), and practitioner perspectives should be taken into account for designing and developing K-12 AI education. Moreover, the two basic curriculum design approaches-process and praxis-describe how to nurture student literacy [20]. Accordingly, the primary goal of this study is to propose a framework for teachers to promote AI literacy and competency. This comprehensive framework explains core knowledge (the content) and pedagogy (the process of learning) that help students improve their AI literacy and competency (learning outcomes). In order to accomplish this goal, we defined AI literacy and competency and co-designed the framework with teachers. Since the learning needs of adults who have not received AI education are comparable to those of students, the framework can be applied to both non-AI professionals and K-12 students, i.e., AI for All and AI for K-12. The findings of this study will contribute to AI education by enhancing our knowledge of AI literacy and competency for K-12 students. Researchers and practitioners would use the definition and theoretical framework suggested by the findings to design research and learning activities related to AI literacy and competency.

Literature review

This study’s conceptual framework is comprised of our definition of AI literacy and competency, as well as two approaches to curriculum design: praxis and process. The definition offers suggestions for the content and learning outcomes that should be incorporated in K-12 AI education; the two approaches recommend that effective pedagogies be used. The terms AI literacy and competency were initially defined in this section, along with the two approaches to curriculum design-process and praxis-discussed. Next comes the critical discussion of relevant papers on the AI education framework in order to identify research gaps.

AI literacy and competency

Initially, literacy refers to specific ways of thinking about and performing reading and writing in order to comprehend or express ideas or thoughts in writing within a particular context of use [26]. Digital literacy refers to the ability to appropriately use, assess, and apply digital tools, resources, and services to lifelong learning processes [17]. Clearly, digital competency encompasses more than just proficiency in operating devices and programs; it is also closely intertwined with the ability to communicate using technologies and digital skills. It should include a balanced view of technology for responsible and healthy use of digital technology. Knowledge and attitudes about privacy and security, legal and ethical considerations, and the role of digital technologies in society should be included [17]. Therefore, literacy is directly tied to skills and is more about knowing. Beyond that, competency is the ability to perform a task effectively and successfully [17]. Broadly speaking, literacy is more about knowing, and competency is more about applying the knowledge in an effective and beneficial way. It is related to confidence and attitude, and focuses on how well an AI user does.
Long and Magerko [31] define AI literacy as “a set of competencies that enables individuals to critically evaluate AI technologies, communicate and collaborate effectively with AI, and use AI as a tool online, at home, and in the workplace.” They see this literacy as a set of 17 skills and as an operational definition. AI literacy is clearly related to other literacy such as digital, data, and computational literacy. The relationships could be mutually dependent but exclusive [31]. For example, AI literacy requires users to have a fundamental understanding of how to utilize computers in order to comprehend AI. Therefore, digital literacy, which refers to the ability to use computers to complete a task [33], is a
prerequisite for AI literacy. Considering the close connection between data and machine learning (a branch of AI), data literacy refers to the capacity to understand, work with, evaluate, and argue with data as part of a more comprehensive process of inquiry into the world [45], which largely overlaps with AI literacy. Moreover, the other two literacies, such as computational and scientific, may not closely relate to AI literacy. Computational literacy involves exploring and communicating ideas through code [44]; therefore, it is not necessarily a prerequisite for AI literacy that does not require writing codes to understand how AI works. Similarly, AI literacy does not require scientific literacy, which refers to an appreciation of the nature, contributions, and basic limitations of science [22]. The definition of AI literacy is one of the first for non-AI professionals, which could give educators and researchers new insights about the content and skill assessments related to AI. Its development was based on a literature review done by two engineering professors. Most of the literature was published at engineering conferences around 2018-2019. These imply that this definition is driven by an engineering perspective and higher education. The definition may not be appropriate for education, which is supported by the major limitations of this definition raised by the authors [31]. They noted that AI education is still in its early stages and requires additional empirical studies, especially on teacher perspectives, to get a robust and accurate understanding of AI literacy for a non-technical audience [31]. Moreover, in the definition, the terms of literacy and competence were mixed use in the literature (e.g., [31]). However, as we discussed literacy and competency are related but different in educational research [17]. Therefore, it is necessary to redefine AI literacy and competency.

Process and praxis design approaches in AI education context

Adopting a specific curriculum design strategy has a significant impact on teaching and learning processes [38]. The four basic approaches to curriculum design are content, product, process, and praxis. The content approach views education as knowledge transmission, e.g., a syllabus with defined content [20,27]; the curriculum as product views teaching as grading and focuses on student performance [20,27], i.e., student learning outcomes. It often creates lists of competences, telling students what to learn and how to learn it, and emphasizes education with pre-defined outcomes. In AI education, these two approaches are informed by the definition of AI literacy and competency.
Instead of pre-defined content and outcomes, the process design approach stresses how teachers, students, and content adapt, and it also views teaching as development. Triadic interactions change learning goals [27]. The curriculum is a guide to teaching, not a set of items for instructors to cover and deliver [20]. Content is tailored to the needs and interests of the students; learning outcomes are determined by teachers and students but are not universally applicable. This approach values student-centered learning experiences. To make sense of what is being learned, the praxis design approach places a focus on how it may be used in the real world. Students, under the supervision of their teachers, collaborate on solutions to real-world problems and develop a strategy for learning the necessary material and obtaining the desired outcomes. Both the process of learning and its outcomes are constantly assessed. Therefore, problem-based learning is often adopted. AI education is still new to schools; its process and praxis approaches are not clear and require more empirical research.
Overall, the first two approaches create a set of documents for implementation, and the last two approaches advocate student-centered approaches, shifting the focus of the curriculum from teaching to learning [27]. Therefore, the definitions of AI literacy and competency and teacher perspectives on pedagogies would contribute to the development of an AI education framework for K-12.

Three major frameworks for AI education in K-12

A few key frameworks were suggested in the AI K-12 education
research in the past 5 years. One of the first frameworks, known as “Five Big Ideas in AI,” was suggested by Touretzky et al. [43]. In 2018, there was little external guidance from the literature on the content and scope of AI education for K-12 students [42]. The AI4K12 Steering Committee, which consists of David Touretzky, Christina Gardner-McCune, Fred Martin, and Deborah Seehorn, began their work by coming up with a list. This list serves as the organizing framework for the guidelines, which are developed based on the CSTA Computing Standards. Those standards are structured around the same five core ideas [11]. The five big ideas are:
  • Perception: Computers use sensors to get information about their environment. Understanding what the senses are trying to tell us is what we call perception.
  • Representation and Reasoning: Agents keep models of the world and utilize them to make decisions. Representations are the driving force behind reasoning, and reasoners use them.
  • Learning: Computers can keep learning from data. By modifying the representations within a decision tree or neural network, a machine learning algorithm creates a reasoner.
  • Natural Interaction: To communicate with people in a natural way, intelligent agents need access to a wide range of information. The information includes common sense, culture, human emotions, and knowledge of language.
  • Societal Impact: There will be positive and negative effects of AI on society. The topics include the economic effects of automation, the fairness and transparency of automated decision-making systems, cultural considerations of AI algorithms, and the use of AI for social good.
These big ideas help educators and researchers set the goals of their designed AI education and identify the content needed. Accordingly, this framework used content and product approaches.
The second framework discussed was proposed by Chiu et al. [8] in their project AI4future. Education and engineering professors co-designed the framework with school teachers for middle school students. Fig. 1 shows an infographic presenting their proposed AI curriculum. In the core of the circle, the curriculum begins with an introduction to AI-big data, machine learning, and cloud computing. Another major focus is on ethical issues in the usage of AI applications as well as their societal impact. The middle pink circle depicts our coverage of various branches of AI: perceptual machine intelligence, e.g., “see” and “hear,” human language technologies, e.g., “speak” and “read and write,” integrated intelligences, e.g., machine reasoning, simulation for problem solving, and content creation and generation. The outside green circle depicts many AI-supported applications, many of which have significant societal ramifications, particularly for the future workforce. Similarly, the first framework also focuses on content and product approaches, helping AI educators create their learning and teaching content.
The last framework was drawn from teachers’ perspectives in a school-university partnership project. Chiu [6] added pedagogy to the framework; see Fig. 3. The framework has three layers, shown in light blue, white, and dark blue. The core of the model in light blue shows the three core content components that should be included in AI education for middle school students. The content components are what AI is, how AI processes data, and what impact AI has. The middle layer in white suggests three pedagogies: student relevance, teacher-student communication, and flexibility. They are essential for the effective teaching of AI concepts and knowledge. Students should feel relevant when learning
Fig. 1. Infographic for AI curriculum for K-12 [4].
AI, as it is around students’ lives. Students should use design solutions for authentic problems. In teacher-student communication, teachers should utilize unfamiliar graphics and consistent language to convey AI terminologies and algorithms since they are too new and abstract for young children. This language is more likely to foster teacher-student communication. The last component is flexibility. Teachers can use module and level-up content that provides a learning pathway that directs and guides learning. The last layer in dark blue is the outcome of AI education. The model used the roles of students in an AI-based society. The outcomes include AI users, developers, researchers, and ethical designers. The approach makes the outcomes explicit to students. The process approach was added to the framework.

Research gaps

These three frameworks are significant as they set the content knowledge school students need to learn and suggest pedagogies for teachers. However, they were developed at an early stage [31]. AI learning for K-12 should provide education equality, reducing the digital divide [46]. Literacy (knowing) and competency (how well you do) are related, but different [17]. Literature on AI education interchanges the terms AI literacy and competency (e.g., [31]). We frequently settle with literacy instead of striving for competency since it requires less effort. AI education should focus on AI competency due to its disruptive nature. With the advent of generative AI tools such as ChatGPT and Sora, AI literacy and competency need to be revisited [3]. Self-reflective mindsets and life-long learning skills are important to AI education. Therefore, it is necessary to re-examine these three pieces of work for a more comprehensive framework from the perspective of experienced AI teachers, as teacher perspectives could refine the existing frameworks.

This study and method

Research goal

The two main goals of this paper are to (i) define AI literacy and competency for a non-technical audience, i.e., K-12 education, and (ii) suggest a more comprehensive framework for AI education. How teachers design AI teaching and learning activities directly impacts how and what students learn. Teachers’ perspectives are crucial for making sense of any innovative education [14]. Accordingly, to achieve these two goals, we first presented the definitions of AI literacy and competency and co-designed the framework with experienced AI teachers. The definitions and the curriculum design approaches serve as a lens to reexamine the three major frameworks available in the literature.

The definitions of AI literacy and competency

Long and Magerko’s [31] definition of AI literacy is exceedingly broad, possibly overly complicated, and unsuitable for the K-12 setting. The inclusion of self-reflective mindsets, which are critical for lifelong learning about emerging AI technology, in the definition was missing. AI technologies are expected to stay, evolve, and have an increasing impact on our lives, work, and studies. More students will learn with more advanced AI technologies; more employees will work with the technologies; and more people will live with the technologies. To succeed in the AI age, students must be able to continuously evaluate their own understanding of AI and stay up-to-date on its advancements. This is in line with the recommendations made by UNESCO about AI competency for educators and students: “The knowledge, skills, and attitudes students should acquire to understand and actively engage with AI in a safe and ethical manner in school and beyond.”.
This study revised the definition of AI literacy to improve its suitability for K-12 education and defined AI competency by adding the aspects of student confidence and self-reflective mindsets to AI literacy,

as follows:

  • AI literacy is defined as “an individual’s ability to clearly explain how AI technologies work and impact society, as well as to use them in an ethical and responsible manner and to effectively communicate and collaborate with them in any setting. It focuses on knowing (i.e. knowledge and skills).”.
  • AI competency is defined as “an individual’s confidence and ability to clearly explain how AI technologies work and impact society, as well as to use them in an ethical and responsible manner and to effectively communicate and collaborate with them in any setting. They should have the confidence and ability to self-reflect on their AI understanding for further learning. It focuses on how well individuals use AI in beneficial ways.”.
In this paper, we used technology, impact, ethics, collaboration, and self-reflection to label the confidences and abilities in the definition of AI competency to analyze the data collected, as follows:
  • Technology: confidence and ability to clearly explain how AI technologies work
  • Impact: confidence and ability to clearly explain how AI technologies impact society
  • Ethics: confidence and ability to use AI technologies in an ethical and responsible manner
  • Collaboration: confidence and ability to effectively communicate and collaborate with AI technologies in any setting
  • Self-reflection: confidence and ability to self-reflect on their AI understanding for further learning. Individuals with stronger selfreflective mindsets are more likely to keep reviewing their AI knowledge and identify areas and needs for further learning.

Participants

There were a total of 30 teacher participants involved in this study, split evenly between 15 Hong Kong middle schools ( 2 teachers per school). The teachers’ average age was 32 , and there were 24 male and 6 female participants. All of them had at least three years of experience teaching AI. Their academic backgrounds included computer science, mathematics, the sciences, business, design, and technology. Each district had its own set of schools, which ranged in academic quality and social status. Another key participant was a researcher who is a professoriate staff member with an academic background in STEM education, particularly in mathematics and technology. The researcher has five years of experience promoting AI education in middle schools through the creation of learning materials, testing their effectiveness, and providing teacher training. By considering that AI education for K-12 has started in 2019, the participants were experienced.

Research design and procedure

This study used a co-design approach to propose the framework. The researcher was involved in the design processes and worked together with the teachers. An iterative co-design cycle involving data collection, analysis, and design was conducted. The researcher and teachers contributed to the development of the comprehensive framework through the definitions, design approaches, and three existing major frameworks discussed in the literature review.
The main researcher began running a 3-hour workshop on the four approaches, the three major frameworks, and the definitions of AI literacy and competency for all the teachers. There were four cycles of data collection. In each cycle, the researcher conducted a meeting with the teachers and collected their comments. The teachers were divided into five groups of six and drafted and revised the framework and learning experiences from the previous cycle (the first cycle used the three major frameworks discussed in the review). Each meeting lasted for 4 h with a
15 -minute break. Thus, the researcher and the participants jointly proposed how AI education should be designed and delivered, and the teachers used the materials as evidence to support their ideas.

Analyses, reliability and validity

In each of the meetings (i.e., cycles), all the participants co-designed the framework and learning experience with the researcher. At the end of each cycle, the teacher participants voted on the changes to the framework and learning experiences using an immediate response digital platform. We used a 75 percent agreement (consensus) level, aligned with previous similar studies [41]. This level indicates the high reliability of the analysis.
We used six aspects to discuss the validity of this qualitative design and analysis [15,30]. Appropriateness: First, we used the three major existing AI education frameworks in the literature. Second, the main researcher has designed and delivered AI education for K-12 since 2018 and published more than 15 journal papers related to AI education. The participants were experienced AI education teachers. These show that the methods and materials used for gathering data were suitable for the context and research question. Triangulation: the teachers with different teaching subjects and schools were recruited to avoid individual (subject) biases and provide a more diverse view. Credibility: The credibility of qualitative research is determined by the researcher’s competence in the research process. The authors had experience researching, designing, and teaching AI in different regions. Ethical procedures: The corresponding authors’ university granted ethical permission for this work, addressing the moral implications of this date collection. All the participants understood their rights and gave consent. Transferability: the participants’ major teaching subjects were diverse, which provides a wide possible range of data. Respondent validation: this study was supported by the three major AI education frameworks; thereby, it is credible and valid.

Results and discussions

The confidences and abilities and four curriculum design approaches were employed as a conceptual framework in this study to analyze the data obtained in each cycle. The first two meetings have identified the five key components of the proposed comprehensive framework; see Fig. 3. They are technology, impact, ethics, collaboration, and selfreflection. Each of the findings is explained in detail below.

The five components in the comprehensive framework

Technology

The first component concerns the core knowledge of AI, which is comparable to the two frameworks proposed by Five Big Ideas in AI and AI4future [4,43]. Content is critical for the design of universal education, i.e., K-12. K-12 education, unlike higher education, requires boundaries. Different schools in the same region must work on the same learning objectives. According to the teachers, core knowledge includes three topics: basic components, perception, and applications.
  • In the basic components, the definition, history, and development of AI should be taught in schools [49]. The following definition of AI was agreed upon by all the participants: “AI refers to a machine’s ability to do tasks equivalent to human learning and decision-making.”. This definition is consistent with the findings of Chiu et al. [8] and Touretzky et al. [42]. The students must understand essential concepts such as big data, machine learning, and cloud computing. They must understand the five primary and inherent characteristics of big data, which are velocity, volume, value, variety, and veracity [19]. To properly understand how AI machines use data to enhance their skills, the topic of machine learning should encompass training models and learning algorithms [4,10]. Cloud computing is required
    for huge data processing in order to better train models and/or algorithms. Furthermore, the history and development of AI are critical topics in K-12. All the participants agreed that the students should understand AI history as well as contemporary advancements such as the “fourth industrial revolution” and generative AI, e.g., ChatGPT and Sora [12,49]. The students must comprehend how AI machines differ from non-AI machines. Non-AI machines, for example, have been designed to answer our problems by applying rules or algorithms, whereas AI machines use data to develop and regenerate the rules or models. They should understand that AI is changing the fundamental concept of how machines work and that “data are the new code” [49]. The students who have mastered the essential knowledge, in particular, should be able to recognize if the technologies they are employing are AI and comprehend the ramifications of this. They should also be required to describe what types of data AI collects, how AI analyzes data, and how AI learns from data.
  • Perception is the second topic in core knowledge, which is consistent with the research of Chiu et al. [8] and Touretzky et al. [42]. The teachers indicated that the concept of perception in Touretzky et al. study [42] is too abstract. The students did not understand what this word meant. They chose human sensor concepts as perception subtopics. Human sensors-see, read, and write; speak and hear; think and create; reason and simulate-are terms that reflect the definition of AI. Because these terminologies are not overly technical, both students and teachers will understand what they need to learn or teach. The students should understand how each sensor collects and processes data.
  • The third topic is AI applications. The teachers highlighted that the topic’s breadth is more significant than its depth. The students should have a thorough understanding of AI applications, which should include most industries or aspects of daily life such as healthcare, entertainment, transport, logistics, etc. They should employ perception to describe how each application functions, as well as machine learning to construct and develop their own AI applications.

Impact

The second key component suggests three topics that foster student confidence and ability to clearly explain AI technologies impact on society: future of work, social good, and risks.
  • On the topic of the future of work, young children place a high value on their future studies and employment. They need to understand that more future vocations demand AI literacy and that they are increasingly likely to learn with AI in their lives and work with AI in their jobs.
  • The second subtopic is AI for social good. AI can have both positive and harmful effects on society. According to the teachers, the students should learn how AI solves complicated issues by addressing critical social, environmental, and public health concerns [23]. Other than the benefits AI brings to society, students should also learn the risks.
  • The last topic is risk. All the participants agreed that the students should be aware of the potential risks associated with AI. They should investigate how emerging technologies cause trouble and harm in various contexts. These are aligned with the two frameworks [4,43] and other related studies [47,50].

Ethics

The third component is ethics, and there are concerns about AI ethics and human bias. The teachers discussed which concepts should be included in K-12 education. Existing principles or policies published for the public are very complicated. Some of them were used for legal and business purposes. They expressed that “the AI policies include 20 areas for ethical use.” They are too complicated for young children or non-
technical audiences. The teachers finally used IBM AI ethical principles as a starting point and reached a consensus on what to include in this topic. They chose five of them because they are more relevant to school students: fairness and biases; trust and transparency; accountability; social benefit; and privacy and security [18].

Collaboration

The fourth component is to foster student confidence and ability to communicate and collaborate with AI technologies in any setting. The teacher participants pointed out that these confidence and ability were associated with the three themes discussed earlier: technology, impact, and ethics. They suggested that “the students need to learn how to prompt effectively when using ChatGPT,” “the students should examine the reliability of the prediction of an AI prediction system,” “the students should be aware of their privacy rights when using AI systems,” and “the students should be aware that AI systems may collect their data, such as locations and voices.” This is likely to be fostered through case studies, project-based learning, and hands-on activities. Overall, this is aligned with some studies about prompt engineering, which is the technique of structuring text so that a generative AI model can comprehend and understand it. The promoter is a natural language text that enables more appropriate responses from the model [32,35].

Self-reflection

The last key component concerns student self-reflective mindsets that are strongly associated with confidence. A confident student is more inclined to consistently self-reflect on their AI knowledge compared to a less confident learner. Since AI is an emerging and disruptive technology, it is critical that the students constantly evaluate their knowledge of AI in order to stay up-to-date. This is cognitive but affected by its affective or emotional aspects, since cognitive engagement is associated with emotional engagement [4]. For example, students who are more enthusiastic about AI or technology are more likely to read articles and watch videos about the latest developments in AI-related technology [2, 50]. This is very important for lifelong learning skills in an AI-based society.
The last two meetings in the study focused on discussing the process and praxis-based design approaches. The teachers reached a consensus that five essential learning experiences should be taken into account when designing AI teaching and learning activities.

Five essential learning experiences

Five learning experiences for AI education were identified in the meetings. All the teachers agreed that experience is essential for AI learning. The following discusses how learning experiences promote inclusive and diverse AI education.
  • Community engagement: According to Cooper [13] and Mooney & Edwards [36], integrating student learning with the community is a purposeful pedagogical strategy used by instructors to make a connection between what is taught in the classroom and the students’ local communities. As a result, students can apply their ideas based on personal observation and social interaction to design and find solutions to real-world problems in a community. They are more inclined to invest more in learning since the challenges are more relevant and encourage student participation. This community involvement will increase students’ interest and enthusiasm , 28], resulting in more inclusive and diverse AI education [46]. This community involvement will promote AI for social good while also cultivating students’ positive attitudes toward AI. Furthermore, the teachers stated that the design thinking approach-empathize, define, ideate, prototype, and test-should be used to solve community problems. To have a greater impact, community participation may learn a lot from design thinking [4].
  • Global and local case studies: One of the effective teaching methods proposed by the teachers is using media articles to educate about AI ethics and impact. School students can read various web articles, yet there are many biased options or fake news [16]. They are less mature and less capable of judging the reliability of the articles. Unlike ten years ago, more newspaper stories are being published as a result of AI advancement. They have recently become more accessible to both teachers and students. This method creates non-textbook learning, making it more authentic and relevant. Students can more actively reflect on ethical issues raised in newspaper stories [40]. Furthermore, this teaching style can be utilized to teach ethical principles while emphasizing the importance of data sources. This can be used by teachers to illustrate concepts such as fairness and bias, trust and transparency, and accountability.
  • Hands-on activities: Students are surrounded by AI, but not so much in their classrooms. Hands-on activities could help students learn perception better because they put them more in the driver’s seat through physical activity and active learning [48]. Physically active learning can support students’ confidence and ability to model the world and generate creative ideas [48]. These connections may allow multipart tool design and use. Students can solve problems they cannot solve orally or visually by modeling physical systems with their hands [46]. Asking students to use their hands to model physical phenomena using AI will develop their conceptual understanding of AI perception. The teachers also suggested some practical ideas: students could investigate how AI can collect, understand, and identify images, i.e., learn how computers see images. Students can understand what they read and write since AI understands languages and text, and they can develop text analyzers to detect text moods. Furthermore, most students believe AI and robotics are the same technology. Students should learn how AI differs from robotics by doing a hands-on project in which they build a robot with reasoning and perception. According to the teachers, ethical principles, which were often taught using case-study approaches, can be taught through hands-on activities. Students could design one or more biased AI apps and explain how the dataset selection resulted in the biased results. As a result, hands-on activities can better cater to students with individual and learning differences and promote inclusive and diverse AI education [46].
  • Exhibitions and presentations: This strategy is prevalent in the dissemination of student work, particularly in project-based learning. It allows students to consolidate their learning by communicating their processes, thinking about their products, and reflecting on their answers [21]. Teachers indicated in this framework that students should display their learning process and project work. The teacher highlighted in AI education that students should use core knowledge as criteria to present their work in the exhibition. For example, how did they acquire or create the data for building the perception? Are the data ethical? Is their use moral? What is the social impact of their work? What industry will acquire their products? Accordingly, presenting their project work encourages students to tap into their individual differences.
  • Cultural learning: Human values and culture should be considered when learning AI. Data may be biased due to how it is obtained or chosen for usage. Human values and culture play a significant role in this. The answer to the questionAI bias or not”” may not be determined by AI knowledge but by human values and culture. These two elements must be considered by students while learning AI. For example, when developing AI projects to help the elderly, students are obliged to respect their culture and values (e.g., food and diets) but not change theirs. When discussing ethical issues related to driverless cars, students must be aware of local laws and religious beliefs (hurting cows in India may result in serious crimes).
Overall, the teachers’ recommendations for five essential learning experiences would enhance inclusive and diverse AI education through
social good (community participation), students’ relevancy (newspaper articles), active learning (hands-on activities), collaboration, and communication (presentation) (Fig. 2).

Future research directions and recommendations

Compared to the existing major frameworks for AI K-12 education, this framework has two newly added aspects-confidence and selfreflective mindsets-and two newly added abilities: (i) collaboration: ability to effectively communicate and collaborate with AI technologies in any setting; (ii) self-reflection: ability to self-reflect their AI understanding for further learning. This paper focuses on discussing five future research directions to address these confidences and abilities.
  • Prompt engineering: In this new technique, you may rephrase a query, select a style, provide more contextual information, or assign a role to AI when interacting with . The current studies on this technique are restricted to ChatGPT. This is also related to our questioning skills. Most of the young children’s questioning skills are developed in K-12 education. More studies should be conducted to understand how K-12 students interact with AI, i.e., prompt engineering or questioning skills for AI.
  • Data science education: how to communicate with AI is associated with who to use and interpret data. Data literacy is strongly associated with AI literacy [31]. Data science education and data literacy may need to be included in K-12 education. Currently, this literacy is embedded in mathematics or AI subjects. More studies should be conducted to investigate how data literacy relates to AI literacy development.
  • Algorithmic Literacy: To fully communicate with AI, algorithmic literacy may be needed. This literacy includes awareness and knowledge of algorithms, trust and confidence in algorithms, algorithm appreciation, and algorithm avoidance [25] This framework did not discuss this literacy. Future studies should be conducted to define algorithmic literacy and how it affects AI literacy.
  • Self-reflective mindsets: this mindset is understudied in AI education. AI is different from other typical subjects, such as mathematics and sciences in K-12, and is emerging. Social cognitive theory is related to emotional engagement, and self-regulated learning concerns self-reflection. Future studies are recommended to use social cognitive theory and self-regulated learning to identify effective pedagogies to nurture students’ self-reflective mindsets.
  • Empirical research: AI education is still new; more empirical studies should be conducted to refine the definition of AI literacy and competency and the AI education framework.

Conclusion and limitations

Designing an effective AI education in K-12 appears to be quite difficult. A few frameworks-“Five Big Ideas in AI,” AI4future, and a holistic approach-were presented in the literature to help teachers create their curriculum. They do, however, concentrate on recognizing AI content, tools, technical abilities, and thinking. They may not properly provide them with a fundamental understanding and get them ready for a future with AI. To address this issue, this study defined AI competency by adding the aspects of confidence and self-reflective mindsets to AI literacy and proposed a comprehensive framework for AI education in . We believe that the framework promotes inclusive and diverse education, ensuring success for each student [39]. Teachers can design AI education to address students’ needs [37].
We acknowledge two limitations in this study. First, the framework has not been put through its paces in the field, so more research is needed to investigate its usefulness. One way to do this would be to have a variety of schools (including those serving students with special needs) use the framework as a guide for developing and executing AI education initiatives. Second, this study did not address how teacher capacity affects learning design and development, which is a major limitation given that most teachers do not study AI topics in their education and undergo formal AI training. Learning how to design and implement AI teacher professional programs and the impact they have on learning design is an important area for future research.

CRediT authorship contribution statement

Thomas K.F. Chiu: Writing – review & editing, Writing – original draft, Validation, Methodology, Investigation, Funding acquisition, Formal analysis, Conceptualization. Zubair Ahmad: Writing – review & editing. Murod Ismailov: Writing – review & editing. Ismaila Temitayo Sanusi: Writing – review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Fig. 2. A framework for designing a holistic approach for K-12 AI education (adopted from [5]).
Fig. 3. The five key components in the comprehensive framework.

Data availability

The datasets used for the current study are available from the corresponding author on reasonable request.

Funding

This is supported by General Research Fund (project code: 14610522), Research Grant Council, Hong Kong, SAR.

Acknowledgements

The authors thank you for the school teachers for participate this study

References

[1] Barton AC, Tan E, Greenberg D. The makerspace movement: Sites of possibilities for equitable opportunities to engage underrepresented youth in STEM. Teach Coll Rec 2017;119(6):1-44. https://doi.org/10.1177/016146811711900608.
[2] Chiu TKF. Using Self-determination Theory (SDT) to explain student STEM interest and identity development. Instr Sci 2024;58:89-107. https://doi.org/10.1007/ s11251-023-09642-8.
[3] Chiu T.K.F. (2023). The impact of Generative AI (GenAI) on practices, policies and research direction in education: A case of ChatGPT and Midjourney, Interactive Learning Environments, Advanced online publication. https://dx.doi.org/10.1080/ 10494820.2023.2253861.
[4] Chiu TKF. Applying the Self-determination Theory (SDT) to explain student engagement in online learning during the COVID-19 pandemic. J Res Technol Edu 2022;54(1):14-30. https://doi.org/10.1080/15391523.2021.1891998.
[5] Chiu TKF. A holistic approach to Artificial Intelligence (AI) curriculum for K-12 schools. TechTrends 2021;65:796-807. https://doi.org/10.1007/s11528-021-00637-1.
[6] Chiu TKF, Chai CS, Williams J, Lin TJ. Teacher professional development on Selfdetermination Theory-based design thinking in STEM education. Edu Technol Soc 2021;24(4):153-65. https://www.jstor.org/stable/48629252.
[7] Chiu TKF, Ismailov M, Zhou X-Y, Xia Q, Au D, Chai CS. Using Self-Determination Theory to explain how community-based learning fosters student interest and identity in integrated STEM education. Int J Sci Math Educ 2023;21:109-30. https://doi.org/10.1007/s10763-023-10382-x.
[8] Chiu TKF, Meng H, Chai CS, King I, Wong S, Yeung Y. Creation and evaluation of a pre-tertiary artificial intelligence (AI) curriculum. IEEE Transac Edu 2022;65(1): 30-9. https://doi.org/10.1109/TE.2021.3085878.
[9] Chiu, T.K.F., Moorhouse, B.L., Chai, C.S., & Ismailov M. (2023). Teacher support and student motivation to learn with artificial intelligence (AI) chatbot, Interactive Learning Environments, Advanced online publication. https://doi.org/10.1080/10 494820.2023.2172044.
[10] Chiu TKF, Xia Q, Zhou X-Y, Chai CS, Cheng M-T. Systematic literature review on opportunities, challenges, and future research recommendations of artificial intelligence in education. Comput Edu: AI 2023;4:100118. https://doi.org/ 10.1016/j.caeai.2022.100118.
[11] CSTA. (2017). Computer Science Teachers Association (CSTA) K-12 Computer Science Standards, Revised 2017. https://www.csteachers.org/page/standards. Accessed 12 Nov 2022.
[12] Cooper G. Examining science education in ChatGPT: An exploratory study of generative artificial intelligence. J Sci Educ Technol 2023;32:444-52. https://doi. org/10.1007/s10956-023-10039-y.
[13] Cooper JE. Strengthening the case for community-based learning in teacher education. J Teach Educ 2007;58(3):245-55. https://doi.org/10.1177/ 0022487107299979.
[14] Cope B, Kalantzis M, Searsmith D, Cope B, Kalantzis M, Searsmith D. Artificial intelligence for education: Knowledge and its assessment in AI-enabled learning ecologies. Educational Philosophy and Theory 2020;53(12):1229-45. https://doi. org/10.1080/00131857.2020.1728732.
[15] Cypress BS. Rigor or reliability and validity in qualitative research: Perspectives, strategies, reconceptualization, and recommendations. Dimensions of Critical Care Nursing 2017;36(4):253-63. https://doi.org/10.1097/DCC.0000000000000253.
[16] Datta P, Whitmore M, Nwankpa JK. A perfect storm: social media news, psychological biases, and AI. Digital Threats: Research and Practice 2021;2(2): 1-21. https://doi.org/10.1145/3428157.
[17] Falloon G. From digital literacy to digital competence: the teacher digital competency (TDC) framework. Educational Technology Research and Development 2020;68:2449-72. https://doi.org/10.1007/s11423-020-09767-4.
[18] Gerke S, Minssen T, Cohen G. Ethical and legal challenges of artificial intelligencedriven healthcare. Artificial intelligence in healthcare. Academic Press; 2020. p. 295-336. https://doi.org/10.1016/B978-0-12-818438-7.00012-5.
[19] Ghasemaghaei M. Understanding the impact of big data on firm performance: The necessity of conceptually differentiating among big data characteristics. Int J Inf Manage 2021;57:102055. https://doi.org/10.1016/j.ijinfomgt.2019.102055.
[20] Glatthorn AA, Boschee F, Whitehead BM, Boschee BF. Curriculum leadership: strategies for development and implementation. London: SAGE; 2018.
[21] Grant MM, Branch RM. Project-based learning in a middle school: Tracing abilities through the artifacts of learning. J Res Tech Edu 2005;38(1):65-98. https://doi. org/10.1080/15391523.2005.10782450.
[22] Holbrook J, Rannikmae M. The meaning of scientific literacy. Int J Environ Sci Edu 2009;4(3):275-88.
[23] Holzmeyer C. Beyond ‘AI for Social Good'(AI4SG): social transformations-Not tech-fixes-For health equity. Interdisciplinary Science Reviews 2021;46(1-2): 94-125. https://doi.org/10.1080/03080188.2020.1840221.
[24] Hornberger M, Bewersdorff A, Nerdel C. What do university students know about Artificial Intelligence? Development and validation of an AI literacy test. Computers and Education: Artificial Intelligence 2023;5:100165. https://doi.org/ 10.1016/j.caeai.2023.100165.
[25] Du YR. Personalization, Echo Chambers, News Literacy, and Algorithmic Literacy: A Qualitative Study of AI-Powered News App Users. J Broadcast Electron Media 2023;67(3):246-73. https://doi.org/10.1080/08838151.2023.2182787.
[26] Keefe EB, Copeland SR. What is literacy? The power of a definition. Research and Practice for Persons with Severe Disabilities 2011;36(3-4):92-9. https://doi.org/ 10.2511/027494811800824507.
[27] Kelly AV. The curriculum: theory and practice. 6th ed. London: Sage; 2009.
[28] King NS, Pringle RM. Black girls speak STEM: Counterstories of informal and formal learning experiences. J Res Sci Teach 2019;56(5):539-69. https://doi.org/ 10.1002/tea.21513.
[29] Laupichler MC, Aster A, Raupach T. Delphi study for the development and preliminary validation of an item set for the assessment of non-experts’ AI literacy. Computers and Education: Artificial Intelligence 2023;4:100126. https://doi.org/ 10.1075/idj.23.1.03dighttps://doi.org/10.1016/j.caeai.2023.100126.
[30] Leung L. Validity, reliability, and generalizability in qualitative research. J Family Med Prim Care 2015;4(3):324. https://doi.org/10.4103/2249-4863.161306.
[31] Long D, Magerko B. What is AI literacy? Competencies and design considerations. In: Proceedings of the 2020 CHI conference on human factors in computing systems; 2020. p. 1-16. https://doi.org/10.1145/3313831.3376727.
[32] Liu P, Yuan W, Fu J, Jiang Z, Hayashi H, Neubig G. Pre-train, prompt, and predict: A systematic survey of prompting methods in natural language processing. ACM Comput Surv 2023;55(9):1-35.
[33] Martin A, Grudziecki J. DigEuLit: Concepts and tools for digital literacy development. Innovation in Teaching and Learning in Information and Computer Sciences 2006;5(4):249-67. https://doi.org/10.11120/ital.2006.05040249.
[34] Marsh CJ, Willis G. Curriculum: alternative approaches, ongoing issues. Upper Saddle River, NJ: Merrill/Prentice Hall; 2003.
[35] Meskó B. Prompt engineering as an important emerging skill for medical professionals: tutorial. J Med Internet Res 2023;25:e50638. https://doi.org/ 10.2196/50638.
[36] Mooney LA, Edwards B. Experiential learning in sociology: Service learning and other community-based learning initiatives. Teach Sociol 2001;29(2):181-94. https://doi.org/10.2307/1318716.
[37] Priestley M. Whatever happened to curriculum theory? Critical realism and curriculum change. Pedagogy, Culture & Society 2011;19(2):221-37. https://doi. org/10.1080/14681366.2011.582258.
[38] Priestley M, Biesta G, editors. Reinventing the curriculum: new trends in curriculum policy and practice. A&C Black; 2013.
[39] Simmons J, MacLean J. Physical education teachers’ perceptions of factors that inhibit and facilitate the enactment of curriculum change in a high-stakes exam climate. Sport Educ Soc 2018;23(2):186-202. https://doi.org/10.1080/ 13573322.2016.1155444.
[40] Tal T. Pre-service teachers’ reflections on awareness and knowledge following active learning in environmental education. International Research in Geographical and Environmental Education 2010;19(4):263-76. https://doi.org/ 10.1080/10382046.2010.519146.
[41] Teixeira PJ, Marques MM, Silva MN, Brunet J, Duda J, Haerens L, La Guardia J, Lindwall M, Lonsdale C, Markland D. A Classification of Motivation and Behavior Change Techniques Used in Self-Determination Theory-Based Interventions in Health Contexts. Motiv Sci 2020;6(4):438-55. https://doi.org/10.1037/ mot0000172.
[42] Touretzky DS, Gardner-McCune C, Martin F, Seehorn D. Envisioning AI for K-12: What should every child know about AI?. In: Proceedings of AAAI-19; 2019. https://doi.org/10.1609/aaai.v33i01.33019795.
[43] Touretzky D, Gardner-McCune C, Seehorn D. Machine learning and the five big ideas in AI. Int J Artif Intell Educ 2023;33(2):233-66. https://doi.org/10.1007/ s40593-022-00314-1.
[44] Tsarava K, Moeller K, Román-González M, Golle J, Leifheit L, Butz MV, Ninaus M. A cognitive definition of computational thinking in primary education. Comput Educ 2022;179:104425. https://doi.org/10.1016/j.compedu.2021.104425.
[45] Wolff A, Gooch D, Montaner JJC, Rashid U, Kortuem G. Creating an understanding of data literacy for a data-driven society. J Commun Inform 2016;12(3). https:// doi.org/10.15353/joci.v12i3.3275.
[46] Xia Q, Chiu TKF, Lee M, Temitayo I, Dai Y, Chai CS. A self-determination theory design approach for inclusive and diverse artificial intelligence (AI) K-12 education. Comput Educ 2022;189:104582. https://doi.org/10.1016/j. compedu.2022.104582.
[47] Williams R, Ali S, Devasia N, DiPaola D, Hong J, Kaputsos SP, Breazeal C. AI+ ethics curricula for middle school youth: Lessons learned from three project-based curricula. Int J Artif Intell Educ 2023;33(2):325-83. https://doi.org/10.1007/ s40593-022-00298-y.
[48] Yannier N, Hudson SE, Koedinger KR, Hirsh-Pasek K, Golinkoff RM, Munakata Y, Brownell SE. Active learning:”Hands-on” meets “minds-on. Science (1979) 2021; 374(6563):26-30. https://doi.org/10.1126/science.abj9957.
[49] Zhang C, Lu Y. Study on artificial intelligence: The state of the art and future prospects. J Ind Inf Integr 2021;23:100224. https://doi.org/10.1016/j. jii.2021.100224.
[50] Zhang H, Lee I, Ali S, DiPaola D, Cheng Y, Breazeal C. Integrating ethics and career futures with technical learning to promote AI literacy for middle school students: An exploratory study. Int J Artif Intell Educ 2023;33(2):290-324. https://doi.org/ 10.1007/s40593-022-00293-3.

    • Corresponding author at: Faculty of Education, The Chinese University of Hong Kong, Shatin, Hong Kong.
    E-mail addresses: tchiu@cuhk.edu.hk (T.K.F. Chiu), zubairtarar@qu.edu.qa (Z. Ahmad), ismailov.murod.gm@u.tsukuba.ac.jp (M. Ismailov), ismaila.sanusi@uef.fi (I.T. Sanusi).