Earticle

본고는 예비초등교사들의 단소 실기 수업에서 활용할 수 있는 AI 기반 자기성찰 도구를 설계 및 개발하고, 실제 활용 가능성을 탐색하는 데에 목적이 있다. 본 연구는 Richey & Klein(2014)이 제안한 설계ㆍ개발 연구 방법 중 산출물 도구 연구 방법을 토대로 진행되었으 며, 연구 절차는 연구 설계를 위한 이론적 고찰, 도구의 설계 및 개발, 전문가 사용성 평가, 도구의 사용 및 현장 평가의 단계로 구성하였다. 도구는 바이브 코딩 기반 서비스인 래플릿 (Replit)을 활용하여 프로토타입으로 개발하였으며, ChatGPT 5.0 API를 연동하여 AI가 학습 자의 성찰 내용을 분석하고 피드백을 생성할 수 있도록 설계하였다. 이후 음악ㆍ국악실기 및 음악ㆍ국악교육을 전공한 전문가 8인을 대상으로 두 차례의 전문가 사용성 평가를 실시하여 도구 구성, 편리성, 사용 용이성, 교육적 효과, 디자인을 검토하였다. 현장 평가에서는 최종 개발된 도구를 수업 및 개인 연습에서 2주 동안 최소 4회 이상 활용하도록 하였다. 연구 결 과, 본 도구의 편리성과 사용 용이성 측면에서 긍정적인 평가를 받았으며, 교육적 활용 가능 성에 대한 높은 잠재력을 확인하였다. 향후에는 AI 피드백의 정교화나 예비초등교사의 실기 능력과 관련된 다양한 도구 개발이 필요함을 제언하였다.

This study aims to develop an AI-based self-reflection tool for pre-service elementary teachers in the context of Danso performance instruction and to explore its practical applicability. The study employed the product-oriented tool research method within the design and development research framework proposed by Richey and Klein (2014). The research procedure consisted of four phases: a theoretical review for research design, tool design and development, expert usability evaluation, and implementation and field evaluation. The tool was developed as a prototype using Replit, a cloud-based coding platform for Vibe Coding, and was designed to integrate the ChatGPT 5.0 API so that the AI system could analyze learners’ reflective responses and generate feedback. To examine the tool’s structure, convenience, ease of use, educational effectiveness, and design, two rounds of expert usability evaluation were conducted with eight specialists in Korean traditional music performance and music education. For the field evaluation, the final version of the tool was used at least four times over a two-week period during classes and individual practice sessions. The results indicated that the tool received positive evaluations in terms of convenience and usability, and it demonstrated strong potential for educational application. The findings suggest the need for further refinement of AI-generated feedback and the development of additional tools to support performance competencies of pre-service elementary teachers.

본 연구는 예비 교원의 AI·디지털 도구 활용에 대한 인식과 교수설계 역량에 대한 교육요구도를 분석하여 교원양성 과정의 교육적 개선 방향을 탐색하고자 하였다. 이를 위해 218명의 예비 교원을 대상으로 AI·디지털 도구 활용 인식 및 교수설계 요구를 측정하는 설문조사를 실시하였으며, 수집된 자료는 t-검정, Borich 요구도 분석, Locus for Focus 모델을 적용하여 분석하였다. 연구 결과, 예비 교원들은 AI·디지털 도구 활용 수업에 대해 전반적으로 높은 중요도를 인식하고 있었으나, 모든 항목에서 보유도와 중요도 간 유의미한 차이가 확인되었다. Borich 요구도 분석과 Locus for Focus 분석에서 공통으로 우선순위가 높게 도출된 항목은 수업준비 단계의 학습자 수업이해 수준 파악, 평가도구 개발, 학습자 특성 파악, 수업실행 단계의 개별화지도, 상호작용 및 피드백, 학습자 수행유도, 수업환류 단계의 학습데이터 관리였다. 이러한 결과를 바탕으로 예비 교원 대상 AI·디지털 도구 활용 교수설계 교육의 강화 방향과 교원 양성 과정의 개선 방안을 논의하였다.

This study analyzes pre-service teachers' perceptions and educational needs regarding instructional design using AI·digital tools, aiming to derive implications for teacher training programs. A survey was conducted with 218 pre-service teachers, and the data were analyzed using a paired sample t-test, Borich needs analysis, and the Locus for Focus model. Results showed that pre-service teachers perceived the importance of AI·digital tool-based instruction to be high, with statistically significant differences confirmed between perceived importance and current competence levels across all items. Both analyses identified common high-priority areas: identifying learners' understanding levels, developing assessment tools, and analyzing learner characteristics in the preparation stage; individualized instruction, interaction and feedback, and inducing learner performance in the implementation stage; and managing learning data in the reflection stage. Based on these findings, this study presents critical educational domains to be reinforced in teacher training curricula and discusses directions for improving teacher education programs.

이 연구는 러시아어권 중도입국 학습자의 한국 교과 한국어 학습을 지원하기 위해, 한국어와 러시아어 교과서를 의미 기반으로 연계하는 인공지능 기반의 프로 그램 설계를 목적으로 한다. 이 프로그램은 교과서를 직접 번역하는 방식을 벗어 나 다국어 문장 임베딩을 활용하여 한국과 러시아 교과서의 문장을 동일한 의미 공간에 투영한다. 그리고 사용자가 입력한 검색 대상 용어를 중심으로 한-러 문장 간의 의미적 유사성을 산출함으로써 교과서 간 대응 관계를 탐색한다. 이를 위해 한국 고등학교 물리학Ⅰ 교과서와 러시아 9학년 물리학 교과서를 대상으로 문장 단위 의미 매핑과 단원 간 대응 분석을 수행하였다. 또한 스트림릿(Streamlit) 기반 웹 애플리케이션으로 구현하여 용어 검색 및 단원 검색 기능과 PDF 시각화를 제공 함으로써 학습자의 직관적 탐색을 지원하였다. 결과적으로 이 연구에서 제시한 프로그램은 검색 대상 용어를 중심으로 한 교과 내용 탐색은 효과적으로 지원하였 다. 그러나 한-러 교과서 간 내용 비대칭, PDF 구조와 번역의 불완전성에 의한 한계를 나타냈다. 이러한 연구의 결과는 러시아어권 중도입국 학습자를 위한 교과 한국어 학습에 인공지능 기술을 접목한 도구 사용의 가능성을 제시하였다는 점에서 의의를 지닌다.

This study aims to develop an AI-based learning assistance program that links Korean and Russian textbooks based on meaning to support the learning of Korean subjects by Russian-speaking immigrant students. Rather than directly translating textbooks, this program utilizes multilingual sentence embedding to project sentences from both Korean and Russian textbooks into a common semantic space. It then calculates semantic similarities between sentences based on search terms entered by the user, thereby exploring correspondences between the two textbooks. To achieve this, sentence-level semantic mapping and unit-to-unit correspondence analysis were performed on a Korean high school Physics I textbook and a Russian 9th-grade physics textbook. Furthermore, the program was implemented as a Streamlit-based web application, combining term and unit search functions with PDF visualization to support intuitive learner exploration. While the program effectively supported content exploration based on search terms, limitations were identified due to content asymmetry between the Korean and Russian textbooks, PDF structure, and incomplete translations. These findings are significant in that they suggest the potential of AI-based tools for Korean language learning for Russian-speaking immigrant students.

본 연구는 OECD 2025 Teaching Compass 프레임워크에 근거하여 AI 시대 교사의 협력적 에이전시 역량을 진단할 수 있는 도구를 개발하는 것을 목적으로 한다. 기존의 교사 AI 역량 연구가 도구적 접근에 초점을 두었 다면, Teaching Compass는 AI를 교실 내 새로운 교육주체로 인식하고 교사-AI 간 협력적 관계를 강조하는 패 러다임 전환을 제시한다. 문헌 분석을 통해 Teaching Compass의 핵심 개념을 체계화하고, 4개 영역(교사-AI 협력적 에이전시, AI 활용 전문적 역량, 전문적 정체성 유지, 교육생태계 내 협력 리더십) 22문항의 진단도구를 개발하였다. 본 연구에서 개발된 진단도구는 AI 시대 교사가 기술에 종속되지 않으면서도 AI와 효과적으로 협력 할 수 있는 전문적 역량을 체계적으로 진단하고 개발하는 데 기여할 것으로 기대된다.

This study aims to develop a diagnostic tool for assessing teachers' co-agency competencies in the AI era based on the OECD 2025 Teaching Compass framework. While existing research on teachers' AI competencies has focused on instrumental approaches, the Teaching Compass presents a paradigm shift that recognizes AI as a new educational agent within classrooms and emphasizes collaborative relationships between teachers and AI. Through literature analysis, this study systematized the core concepts of the Teaching Compass and developed a diagnostic tool comprising 22 items across four domains: Teacher-AI Co-agency, AI-related Professional Competencies, Professional Identity Maintenance, and Collaborative Leadership in Learning Ecosystems. The diagnostic tool developed in this study is expected to contribute to systematically diagnosing and developing the professional competencies that enable teachers in the AI era to collaborate effectively with AI while maintaining independence from technological dependency.

본 연구는 초등학교 수업에 사용할 수 있는 인공지능 모델 기반 학습 지원도구 개발을 위하여 디자인 씽킹 을 적용한 래피드 프로토타입 개발 방법론을 제시하였다. 디자인 씽킹을 통해 초등교사들은 현장의 필요 성을 반영하고 해결 가능한 문제로 구체화 하여 Low-fidelity 프로토타입을 만들 수 있다. 만들어진 프로토 타입은 기술개발 담당자에 의해 피드백을 받으며 정교화 된다. 다음으로 최종 개발된 Low-fidelity 프로토타 입은 래피드 프로토타입의 설계, 사용성 평가, 개발, 실행의 순환적 단계를 통하여 구체화되고 고도화된다. 최종적으로 초등교사들이 프로그램 개발 단계에 참여하여 현장의 요구를 반영한 실질적인 학습지원도구를 만들 수 있다.

This study presents a rapid prototype development methodology using design thinking to develop an AI learning support tool for elementary school. Elementary school teachers can participate in the program development phase to create practical learning support tools that reflect the needs of the field. Design thinking will enable elementary school teachers to create a low-fidelity prototype by reflecting the needs of the field and incorporating them into soluble problems. The prototypes are elaborated with feedback from technic development staff. The final low-fidelity prototype is then refined through the cyclical steps of design, usability evaluation, development, and implementation of the rapid prototype.

Deaf and Hard of Hearing (DHH) children face challenges in language acquisition due to hearing loss, which is one of the most crucial factors in language development. As a result, most DHH individuals communicate using ASL gloss instead of conventional sentences used by hearing individuals. However, because they are more accustomed to ASL gloss, they often struggle with constructing and understanding standard sentences, making communication with hearing individuals difficult. To address this issue, this study proposes DHHStoryGen, an automated storybook creation system that supports the language enhancement of DHH children. DHHStoryGen is built upon a fine-tuned LLaMA model for text generation and a fine-tuned Stable Diffusion model for image generation, both trained on custom datasets specifically designed for this purpose. The text dataset is created by translating ASL gloss into natural English sentences, which is used to fine-tune LLaMA to generate storybook content. Meanwhile, the image dataset is used to fine-tune the Stable Diffusion model to generate illustrations that match with each chapter's narrative. By converting ASL gloss into structured text and providing visually engaging images, DHHStoryGen fosters a participatory language learning environment, serving as an assistant tool to enhance the literacy skills of DHH children.

본 연구에서는 알지오매스를 활용한 수학 탐구활동에서 인공지능에 기반한 평가와 피드백을 제공하는 시스템을 개발하고 현장 적용을 실 시하였다. 이때, 탐구활동은 2022 개정 수학과 교육과정의 수학 교과 역량 신장을 목표로 성취기준과 관련하여 지식･이해, 과정･기능, 가치･태도의 세 범주를 통합한 학습과 평가가 이루어질 수 있도록 설계하였다. 과제를 수행하는 과정에서 학생들의 응답에 대한 피드백과 평가는 대규모 언어 모델(Large Language Models, LLM)을 바탕으로 설계했으며, 최종적으로 학생과 교사에게 학습 과정과 결과를 리포트 형식으로 제공하였다. ‘수 학 탐구활동-평가 연계 인공지능 기반 공학 도구’ 개발은 개발 원리 도출, UI/UX 설계, 프로토타입 개발의 순으로 이루어졌다. 프로토타입은 중학교 변화와 관계 영역에서 2종, 기하 영역에서 1종을 개발했으며, 현장 적용에서 학생들은 인공지능 기반 공학 도구에 긍정적인 인식을 나타냈다.

In this study, we developed and implemented a system that provides AI-based assessment and feedback in mathematical inquiry activities using AlgeoMath. The inquiry activities were designed in alignment with the 2022 revised mathematics curriculum to enhance mathematical competencies. The learning and assessment process integrated three categories—knowledge and understanding, processes and skills, and values and attitudes—in relation to the achievement standards. Feedback and assessment of students' responses during problem-solving were designed based on Large Language Models (LLMs). The system ultimately provided both students and teachers with a report summarizing the learning process and outcomes. The development of the AI-Based Technological Tool for Connecting Mathematical Inquiry Activities and Assessment followed a structured process, including deriving development principles, UI/UX design, and prototype development. Three prototypes were developed: two for the Changes and Relationships domain and one for the Geometry domain at the middle school level. In the field implementation, students demonstrated a positive perception of the AI-based technological tool.

This study examined the selective effects of an AI-based speaking tool on Korean EFL university students’ speaking proficiency and self-efficacy using a quasi-experimental mixed-methods design. Eighty-one intermediate-level learners participated in a 10-week intervention (experimental group: n = 41; control group: n = 40). The experimental group used Speak, a mobile application integrating automatic speech recognition (ASR) and AI-mediated speaking interaction, while the control group completed workbook-based speaking preparation. Speaking proficiency was assessed through task-based speaking tests measuring fluency, grammatical accuracy, lexical complexity, and pronunciation, while self-efficacy was measured using an adapted 5-point Likert-scale questionnaire. Reflective journals and semi-structured interviews were also analyzed to explain learners’ perceptions and the divergence between fluency gains and non-significant affective outcomes. Quantitative results showed significant improvement only in fluency, with no significant gains in grammatical accuracy, lexical complexity, pronunciation, or self-efficacy. Qualitative findings indicated that repeated speaking practice and reduced anxiety supported fluency development, whereas limited metalinguistic feedback and reduced authenticity of interaction constrained broader linguistic and affective gains. The findings suggest that AI-based speaking tools are effective for enhancing fluency through repeated oral rehearsal but provide limited support for interaction-dependent speaking development and efficacy construction. The study highlights the importance of integrating AI-mediated speaking practice with teacher feedback and communicative interaction.

The purpose of this study is to devise an effective teaching method to enhance students’ English writing abilities, verifying the educational efficiency and validity of Editpad, an AI-based paraphrasing tool, as a teaching aid in the elective English course “Advanced English Writing.” To examine the effectiveness of Editpad, a statistical analysis was conducted in terms of convenience, satisfaction, usefulness, reliability, and motivation induction. In experiments 1 and 2, participants perceived the concise design that facilitated easy comparison between paraphrased results and original sentences. Positive perceptions were noted for the Smooth mode and synonym features, improved vocabulary, sentence structure skills, writing proficiency, and plagiarism search function through Editpad usage. These participants’ perceptions were considered as the influential factors that ultimately increased interest and motivation for sentence-level as well as paragraph-level learning.

The technological advances in natural language processing is ushering in a new era of augmented interpreting where human interpreters perform their tasks more easily and better with the support of technology. Against this backdrop, the current study is aimed at developing and presenting the prototype of an AI-based computer-assisted interpreting (CAI) tool to support interpreters in processing numbers in real time by the automated recognition and extraction of number information from the source speech. The prototype system was built by using Google Speech-to-Text API and spaCy for automatic speech recognition (ASR) and named entity recognition (NER), respectively, as main components. During a pilot run, the system performed with the precision of 0.82 and recall of 0.90. The prototype system was configured so that its output was presented in three different layouts: only numbers presented in a dialogue box; sentences with number visualization; and real-time transcription followed immediately by fully visualized sentences. It was expected that presenting the full sentences with number visualization would be the optimal option for interpreters due to the avthe context and easy location of numbers in a given speech segment.

Since Traditional Korean medicine (TKM) doctors use various knowledge systems during treatment, diagnosis results may differ for each TKM doctor. However, it is difficult to explain all the reasons for the diagnosis because TKM doctors use both explicit and implicit knowledge. In this study, an upgraded random forest (RF)-based evaluation tool was proposed to extract clinical knowledge of TKM doctors. Also, it was confirmed to what extent the professor's clinical knowledge was delivered to the trainees by using the evaluation tool. The data used to construct the evaluation tool were targeted at 106 people who visited the Sasang Constitutional Department at Kyung Hee University Korean Medicine Hospital at Gangdong. For explicit knowledge extraction, four TKM doctors were asked to express the importance of symptoms as scores. In addition, for implicit knowledge extraction, importance score was confirmed in the RF model that learned the patient's symptoms and the TKM doctor's constitutional determination results. In order to confirm the delivery of clinical knowledge, the similarity of symptoms that professors and trainees consider important when discriminating constitution was calculated using the Jaccard coefficient. As a result of the study, our proposed tool was able to successfully evaluate the clinical knowledge of TKM doctors. Also, it was confirmed that the professor's clinical knowledge was delivered to the trainee. Our tool can be used in various fields such as providing feedback on treatment, education of training TKM doctors, and development of AI in TKM.

Since Traditional Korean medicine (TKM) doctors use various knowledge systems during treatment, diagnosis results may differ for each TKM doctor. However, it is difficult to explain all the reasons for the diagnosis because TKM doctors use both explicit and implicit knowledge. In this study, an upgraded random forest (RF)-based evaluation tool was proposed to extract clinical knowledge of TKM doctors. Also, it was confirmed to what extent the professor's clinical knowledge was delivered to the trainees by using the evaluation tool. The data used to construct the evaluation tool were targeted at 106 people who visited the Sasang Constitutional Department at Kyung Hee University Korean Medicine Hospital at Gangdong. For explicit knowledge extraction, four TKM doctors were asked to express the importance of symptoms as scores. In addition, for implicit knowledge extraction, importance score was confirmed in the RF model that learned the patient's symptoms and the TKM doctor's constitutional determination results. In order to confirm the delivery of clinical knowledge, the similarity of symptoms that professors and trainees consider important when discriminating constitution was calculated using the Jaccard coefficient. As a result of the study, our proposed tool was able to successfully evaluate the clinical knowledge of TKM doctors. Also, it was confirmed that the professor's clinical knowledge was delivered to the trainee. Our tool can be used in various fields such as providing feedback on treatment, education of training TKM doctors, and development of AI in TKM.

This study designs, develops, and validates an AI-based Data-Driven Learning (DDL) web tool for elementary English learners and teachers. While DDL has shown strong pedagogical benefits through corpus-based inductive learning, its use in elementary contexts remains limited due to learners’ low proficiency, cognitive constraints, limited corpus access, and the complexity of existing tools. Meanwhile, artificial intelligence (AI) offers new opportunities but raises concerns regarding reliability, control, and ethics. To address these issues, this study proposes a controlled AI-based DDL tool (AIDDL) that maintains core DDL principles while ensuring learner appropriateness and instructional control. Using a designbased research approach, iterative cycles of design, implementation, analysis, and revision were conducted. AI-DDL was developed based on four principles: exploratory learning support, curriculum alignment, controlled AI use, and learnerfriendly design. It integrates textbook-based corpus data with selectively generated AI content. Key features include concordance exploration, collocation mapping, inference prompts, and storage and review functions. Evaluation by 11 experts showed high design validity (M = 4.69) and usability (M = 4.56). Revisions improved interface clarity, level adjustment, and system stability. The findings imply that AI-DDL is pedagogically sound and practically applicable, offering a viable model for controlled AI integration in elementary language learning

AI 리터러시가 모든 학습자의 보편적 기초소양으로 요구되고 있으나, 초･중등 학생의 AI 리터러시를 체계적으로 측정할 수 있는 현장 적합형 진단도구는 충분히 개발되어 있지 않다. 본 연구는 OECD(2025) AI 리터러시 프레임워크를 이론적 기반으로, 한국 초･중등 교육 현장에 적합한 AI 리터러시 진단도구를 탐색적으로 개발하고 그 활용 가능성을 현장 적용을 통해 검토한 기초 단계 연구이다. 진단도구 개발을 위해 전문가 검토를 실시하였으며, 이후 C지역 디지털 기반 교육혁신 연구학교에 재학 중인 중학생을 대상으로 2025년 8월(사전)과 11월(사후) 온라인 설문을 실시하여 내적 일관성 신뢰도, 대응표본 t검증, 효과크기 분석을 수행하였다. 그 결과, 5개 영역 25개 문항으로 구성된 진단도구가 개발되었으며, 전문가 검토를 통해 문항별 타당도 평균 4.50점, 내용타당도･설명력･유용성 각 4.88점(5점 만점)의 예비적 내용타당도가 확인되었다. 사전검사 전체 참여자(N=2,583) 자료에 기반한 Cronbach's 는 영역별 .867∼.918, 전체 척도 .969로 내적 일관성이 확인되었다. C지역 중학생 대응표본(N=796)에 대한 사전-사후 분석 결과, 5개 전 영역에서 통계적으로 유의미한 AI 리터러시 향상이 확인되었으며(p<.01∼.001), 25개 문항 중 21개(84%)에서 유의한 변화가 나타났다. 다만 본 연구는 확인적 요인분석이나 구조방정식 모형을 통한 심리측정적 타당화에 앞서는 기초 단계 연구로서, 향후 엄밀한 타당화 연구와 효과 검증 연구가 후속될 필요가 있다.

Although AI literacy is increasingly recognized as a fundamental competence for all learners, field-applicable diagnostic tools for systematically measuring K?12 students' AI literacy remain insufficiently developed. This foundational-stage study developed an AI literacy diagnostic tool grounded in the OECD AI Literacy Framework (2025) and explored its applicability through field implementation in Korean primary and secondary education settings. To develop the tool, an expert review was conducted. The tool was then administered to middle school students at digital-based education innovation schools in Region C, with online surveys conducted in August (pre-test) and November (post-test) 2025, from which internal consistency reliability, paired-sample t-tests, and effect size analyses were computed. As a result, a diagnostic tool comprising 25 items across five domains was developed, and the expert review confirmed preliminary content validity, with a mean item-level validity rating of 4.50 and mean ratings of 4.88 for content validity, explanatory power, and usefulness (out of 5). Cronbach's computed from pre-test data (N=2,583) ranged from .867 to .918 across domains, with a total scale of .969, indicating strong internal consistency. Pre- and post-test analysis of matched middle school samples (N=796) confirmed statistically significant improvements across all five AI literacy domains, with 21 of 25 items showing significant change. As a foundational-stage study preceding confirmatory factor analysis or structural equation modeling, subsequent rigorous psychometric validation and controlled effectiveness studies are warranted.

본 연구의 목적은 지능적 문제해결학습(Intelligent Problem-Solving Learning: IPSL)에 근거하여 중등 AI 활용 수업의 질적 특성을 분석할 수 있는 수업분석 도구를 개발하고, 그 내용타당도와 사례 기반 적용 가능성을 검토하는 데 있다. 이를 위해 IPSL 개념모형과 설계 원리, AI 활용 수업 관련 선행연구를 분석하여 5개 영역, 15개 세부 준거, 3수준 루브릭으로 구성된 분석도구 초안을 개발하였다. 이후 전문가 7인의 평정을 통해 I-CVI, S-CVI/Ave, 전문가 간 합의도(rWG)를 산출하고, 서술형 의견을 반영하여 도구를 수정·보완하였다. 또한 국어과 ChatGPT 활용 글쓰기 수업, 역사과 wrtn 활용 관점 탐구 수업, 영어과 AI 튜터 활용 영상 제작 수업에 도구를 예비 적용하여 분석자 간 평정 일관성과 질적 해석 가능성을 검토하였다. 연구 결과, 최종 분석도구는 가치·정체성·공동체 지향성, 메타학습·성찰 구조, 복합·융합 문제해결 설계, 인간–AI 역할 및 협업 구조, 미래지향 역량·전이 가능성의 5개 영역으로 구성되었다. 전문가 검증 결과 전체 도구의 S-CVI/Ave는 .95, 세부 준거 수준의 S-CVI/Ave는 .93으로 나타났으며, 사례 적용 결과 과목별 Krippendorff’s α는 국어과 .813, 역사과 .827, 영어과 .841로 확인되었다. 질적 분석에서는 국어과 사례가 AI 협업형 글쓰기 수업, 역사과 사례가 AI 보조형 역사 관점 탐구 수업, 영어과 사례가 AI 튜터형 영어 산출물 제작 수업의 특성을 보였다. 본 연구는 AI 활용 수업을 AI 사용 여부나 산출물 완성도 중심이 아니라, 학습자의 판단과 성찰, 인간–AI 역할 분담, 전이 가능성 중심으로 분석할 수 있는 준거 기반 루브릭을 제안했다는 점에서 수업성찰, 수업컨설팅, 교사교육에 시사점을 제공한다.

This study developed an instructional analysis tool for examining the qualitative characteristics of secondary AI-integrated instruction from the perspective of Intelligent Problem-Solving Learning (IPSL) and examined its content validity and case-based applicability. Based on the IPSL conceptual model, design principles, and previous studies on AI-integrated instruction, an initial rubric consisting of five domains, fifteen sub-criteria, and three performance levels was developed. Seven experts reviewed the tool, and item-level content validity indices (I-CVI), scale-level content validity indices based on the average method (S-CVI/Ave), and inter-expert agreement (rWG) were calculated. The tool was revised using quantitative results and qualitative comments, and then pilot-tested in three secondary classroom cases: a Korean language writing class using ChatGPT, a history inquiry class using wrtn, and an English video production class using AI tutor tools. The final tool included five domains: value, identity, and community orientation; metalearning and reflective structure; complex and convergent problem-solving design; human?AI role distribution and collaboration structure; and future-oriented competencies and transferability. The expert review supported content validity, with an overall S-CVI/Ave of .95 and an S-CVI/Ave of .93 for the sub-criteria. Krippendorff’s α values were .813, .827, and .841 for the Korean language, history, and English cases, respectively, indicating stable inter-rater consistency. Qualitative analysis showed that the cases represented AI-collaborative writing, AI-assisted historical perspective inquiry, and AI-tutor-based English production classes. These findings suggest that the tool can diagnose AI-integrated instruction by focusing on AI’s instructional role, learners’ judgment and reflection, human?AI role distribution, and transferability, rather than merely on AI use or final products. The study offers implications for lesson reflection, instructional consulting, and teacher education.

본 연구는 ChatGPT를 활용한 글쓰기 의사소통 역량을 평가하는 AI 평가도구 개발에 초점을 두었다. 개발 과정에 있어 내용 전문가와 기술 전문가 간의 상호작용을 통해 IT와 인문사회 영역의 융합과정을 탐구하였다. 내용 전문가와 기술 전문가는 긴밀한 소통 및 상호작용을 통해, 글쓰기 의사소통 역량의 채점기준을 탑재하였고 각 영역별(내용의 적절성, 조직의 효과성, 어법의 정확성) 점수와 피드백을 제공하는 AI 평가도구를 개발하였다. 이 과정에서 본 연구는 내용과 기술이 어떻게 결합하는지에 대한 과정을 드러내고, 이후 생성형 AI를 활용한 평가도구 기술 전문가를 포함한 융합연구자들이 유의해야 할 사항 등에 대해 제시하였다.

This study focused on the development of an AI essay tool for assessing writing-communication competence using ChatGPT. During the development process, the interaction between content expert and technical expert was emphasized to explore the fusion of IT and humanities and social sciences. Through close communication and interaction between the content and technical experts, they incorporated scoring criteria for writing-communication competence and developed an AI essay test tool that provides scores and feedback in the appropriateness of content, effectiveness of organization, and accuracy of grammar. This process revealed how content and technology combine and presented considerations for future fusion researchers, including technical experts in generative AI assessment tools.

게임 그래픽 제작 분야는 기술적 진보가 가속화 되어 가고 있는데 리소스 제작 도구의 다양한 출현과 개발자의 학습과 적절한 사용도 큰 몫을 담당하고 있다. 최근에 빠르게 발전하고 있는 분야인 AI기반의 콘텐츠 제작 도구가 게임 개발의 여러 영역에 다양한 영향을 줄 것으로 보인다. 그 중 비교적 개발 파이프라인에 빠르고 안정적인 도입이 가능한 AI 기반의 텍스쳐 자동 생성 도구를 기존에 사용하고 있는 텍스쳐 제작 도구와 비교 분석 해 보았다. 이를 통해 효과적인 텍스쳐 생성 프로세스 파이프라인 구축을 위해 AI 기반의 텍스쳐 자동 생성 도구의 적극적인 도입의 필요성과 함께 이후 연구의 방향을 제시한다.

본 논문은 AI가 완성된 이미지만을 반환하는 기존 생성형 패러다임의 한계를 넘어, 원격으로 사용자 PC의 그래픽 도구(GIMP)를 직접 조작하는 '도구 구동형 협업 워크플로우'를 제안하고 구현한다. 제안 시스템은 최종 결과물이 아닌, 붓질과 레이어 생성 등 창작 '과정' 자체를 핵심 산출물로 다루어 과정의 보존과 재현성을 목표로 한다. 이를 위해 생성형 AI, MCP(: Model Context Protocol) 기반 브로커, GIMP 플러그인을 결합한 3계층 아키텍처를 설계 및 구현하였다. 중앙 브로커는 멱등성(idempotency)과 순서 보장을 통해 드로잉 명령의 신뢰성 있는 원격 실행을 담당하며, 모든 도구 조작 단계는 상세한 프로비넌스(provenance) 로그로 기록되어 작업의 완전한 재현을 보장한다. 시스템 평가는 명령 왕복 지연(RTT), 신뢰성 등 과정 중심 지표를 통해 수행되었으며, 실험 결과 낮은 지연 시간과 높은 안정성을 확인했다. 본 연구는 AI를 서버에 종속된 이미지 생성기에서 인간과 도구를 공유하는 창작 파트너로 기능할 수 있는 새로운 워크플로우를 제시하며, 이는 예술가의 통제권과 협업 가능성을 탐색하는 기반을 마련한다.

This paper proposes and implements a "tool-driven collaborative workflow" that manipulates the user's PC graphics tool (GIMP) remotely. This workflow transcends the limitations of existing generative paradigms, which only produce finished images. The proposed system treats the creative process itself, such as brushstrokes and layer creation, as the core output rather than the final result. The goal is to preserve and reproduce this process. To accomplish this goal, the paper presents a three-tier architecture that combines generative AI, a Model Context Protocol (MCP)-based broker, and a GIMP plugin. The central broker ensures the reliable, remote execution of drawing commands through idempotency and sequence guarantees. Every tool operation is recorded in a detailed provenance log to guarantee complete reproducibility of the work. The system was evaluated using process-centric metrics, such as command round-trip delay (RTT) and reliability. Experimental results confirmed low latency and high stability. This research presents a new workflow that redefines AI, transforming it from a server-dependent image generator into a creative partner that shares tools with humans. This workflow lays the foundation for exploring the artist's control and collaborative potential.

본 연구는 피드포워드 기반 협력적 쓰기 AI 도구 개발을 제언하는 데 목적이 있다. 기존 쓰기 교육에서 피드백의 한계를 보완한 피드포워드의 개념과 특징을 바탕으로 협력적 쓰기 AI 도구를 구상 및 제언하며, 도구의 실효성을 검증하기 위해 중등 국어 교사 인터뷰를 실시하였다. 피드포워드가 구현되기 위해서는 ‘적극적 상호작용’, ‘적시성’, ‘다층적 피드백’, ‘후속 활동 유발’, ‘메타인지’가 반영되어야 함을 제시하였다. 이에 기반한 협력적 쓰기 AI 도구인 ‘쓰담’을 개발하였으며 현장 전문가 인터뷰를 통해 프로그램의 강점과 약점, 실제 적용 가능성을 파악했다. 학습자의 참여를 이끄는 쓰기 교육을 위하여 피드백에서 피드포워드로의 전환을 주장하며, 이를 협력적 쓰기 AI 도구 개발로 실현할 수 있음을 밝힌다.

This study aims to develop a feedforward-based collaborative writing AI tool. We designed and proposed a collaborative writing AI tool based on the concept and characteristics of feedforward, which complements the limitations of feedback in existing writing education. We interviewed middle school Korean language teachers to verify the effectiveness of the tool. It was suggested that in order for feedforward to be implemented, ‘active interaction,’ ‘timeliness,’ ‘multi-layered feedback,’ ‘inducing follow-up activity,’ and ‘metacognition’ must be reflected. Based on this, we developed ‘Ssdam,’ a collaborative writing AI tool, and identified the strengths, weaknesses, and practical applicability of the program through interviews with field experts. We argue for a shift from feedback to feedforward writing education that encourages learner participation and shows that this can be achieved through the development of collaborative writing AI tools.

본 연구는 UNESCO AI 역량 프레임워크를 기반으로 교사와 학생의 AI 리터러시 검사 도구를 개발하고 타당성을 검증하였다. AI 리터러시를 기술 활용 능력을 넘어 인간중심 사고, AI 윤리, 사회적 책임성을 포함한 광의의 개념으로 재개념화하였으며, 교사와 학생의 AI 리터러시를 교육적 관점에서 통합적으로 접근하였다. 교사용 검사 도구는 인간중심 사고, AI 윤리, AI 기술과 응용, AI 페다고지, 전문성 향상의 5개 영역, 15개 구성요소, 45개 문항으로, 학생용은 인간중심 사고, AI 윤리, AI 기술과 응용, AI 시스템 설계의 4개 영역, 12개 구성요소, 36개 문항으로 구성하였다. 경기도 중·고등학생 4,044명과 교사 772명을 대상으로 모형의 적합도를 확인하고, 모형의 타당도를 검증한 결과 모든 영역의 AVE와 CR 값이 기준치를 초과하여 타당도와 신뢰도를 확보하였다. 이 도구는 교사와 학생의 AI 리터러시를 동일한 이론틀 안에서 진단함으로써, 향후 학교 현장에서 AI 리터러시 교육과 교사 전문성 개발, 관련 정책 수립을 위한 기초자료로 활용될 수 있음을 시사한다. 또한 AI 리터러시 교육의 방향 설정과 맞춤형 교육 프로그램 개발을 위한 기초자료로 활용될 수 있을 것이다.

This study developed and validated an AI literacy assessment tool for teachers and students based on the UNESCO AI Competency Framework. AI literacy was reconceptualized as a broad concept that goes beyond technical proficiency to include human-centered thinking, AI ethics, and social responsibility, and the study took an integrated approach to teachers’ and students’ AI literacy from an educational perspective. The assessment tool for teachers consists of 45 items across 15 components in five domains: human-centered thinking, AI ethics, AI technology and applications, AI pedagogy, and professional development. The tool for students consists of 36 items across 12 components in four domains: human-centered thinking, AI ethics, AI technology and applications, and AI system design. The model’s fit was assessed and its validity verified using a sample of 4,044 middle and high school students and 772 teachers in Gyeonggi Province. The results showed that the AVE and CR values for all domains exceeded the benchmark, confirming the model’s validity and reliability. By assessing the AI literacy of both teachers and students within the same theoretical framework, this tool suggests that it can serve as foundational data for future AI literacy education, teacher professional development, and the formulation of related policies in school settings. Furthermore, it can be utilized as foundational data for setting the direction of AI literacy education and developing customized educational programs.