Earticle

이 연구는 학습자들이 챗GPT를 활용해 스포츠 경험에서 포착할 수 있는 몸적 느낌들을 프롬프트 엔지니 어링을 통해 어떻게 언어로 분석, 통합, 확장해 가는가에 대한 탐색을 연구목적으로 두었다. 연구 방법은 담화분석 의 일환인 텍스트 분석이 적용되었다. 논의된 세부문제는 다음과 같다. 첫째, 스포츠 체험의 본질과 챗GPT 사고 의 기술적 특성이 논거 되었다. 둘째, 프롬프트 과정에서의 의미생성과 텍스트 확장구조가 탐색 되었다. 셋째, 체 화된 글쓰기를 수행할 때 제기될 수 있는 학습적 윤리가 검토되었다. 스포츠 체험을 프롬프트를 통해 의미화시킨 다는 것은 반성적 글쓰기에 대한 실험이라 할 수 있다. 동시에 학습자들이 몸 체험에 대한 의미를 새롭게 확대해 설계해 볼 기회라는 해석도 가능하다. 이 논문의 함의는 바로 여기에 있다.

The purpose of this study is to explore how learners analyze, integrate and expand into language through prompt engineering the bodily feelings, which can be captured from sports experience, by using ChatGPT. The research methodology utilized a discourse analysis approach, involving the analysis of textual data. The specific sub-issues that were discussed are as follows: First, the essence of sports experience and the technological characteristics of ChatGPT thinking were argued. Second, meaning generation and text expansion structures in prompt process were explored. Third, learning ethics that can be presented while performing embodied writing was examined. It may safely be said that giving sports experience meanings by means of prompt is an experiment on reflective writing. At the same time, it is also possible to interpret it as an opportunity for learners to design expanded meanings newly about bodily experience. The implications of this paper lie in these.

최근 생성형 인공지능 기술의 급격한 발전은 교육 분야에 혁신적인 변화를 가져오고 있다. 본 연구는 ChatGPT와 프 롬프트 활용이 SW전공 학생들의 C프로그래밍 학습에 미치는 효과성을 분석하였다. 이를 위해 C프로그래밍 수업을 ChatGPT를 활용한 분반과 전통적인 수업방식의 분반으로 나누고 두 분반 간 학습성과 차이를 통계적으로 분석하였다. ChatGPT 활용 분반에는 ChatGPT 활용 자기 주도 학습의 생소함으로 인한 학습 효율성 저하를 최소화하기 위해 학습 영역별로 설계된 프롬프트를 제공하여 하였으며, ChatGPT와 프롬프트 활용의 효용성을 설문 조사를 통해 평가하였다. 연구결과, ChatGPT와 프롬프트 활용은 통계적으로 유의미하게 학습성과 향상에 기여하였으며, 특히 고난도 학습 영역 에서 긍정적인 영향을 미친 것으로 나타났다. 본 연구는 SW전공 학생들의 프로그래밍 학습에서 생성형 인공지능을 활 용한 자기 주도적 학습의 중요성을 실증적으로 입증했다는 점에서 의의가 있다.

The rapid advancement of generative AI has brought significant changes to education. This study examines the effectiveness of ChatGPT and prompt-based learning in C programming for software (SW) major students. We divi ded the course into two groups: one using ChatGPT and the other following a traditional approach, then statistically analyzed their learning outcomes. To mitigate inefficiencies from unfamiliarity with self-directed ChatGPT learning, structured prompts were provided. Survey results indicate that ChatGPT had a positive impact, especially in high-di fficulty areas, enhancing learning efficiency. This study highlights the importance of generative AI in fostering selfdirected programming education for SW students. This study is significant in that it empirically demonstrates the i mportance of self-directed learning using generative AI in programming education for SW-major students.

본 연구는 디자이너와 생성형 인공지능의 협업 디자인에 중점을 두고, 디자인 의도에 부합하는 이미지를 창출하는 것을 목표로 한다. 이론적 배경 부분에서는 제4차 산업혁명과 인공지능 발전의 이론적 기초를 연구하고, ChatGPT와 Midjourney의 작동 원리 및 기능적 특징을 상세히 분석하였다. 이러한 두 도구를 합리적으로 결합하면 이미지 디자인의 적합성과 창의성 수준을 크게 향상시킬 수 있음을 밝혀냈다. 인지학 관점에서 디자이너의 디자인 사고 과정을 분석하고, 본 연구의 디자인 사고 과정을 도출하여 실험에 대한 지침을 제공하였다. 실험 결과, 명확한 디자인 의도 지침 하에서 ChatGPT와 Midjourney가 생성한 이미지는 디자이너의 기대에 부합하는 것으로 나타났다. 이는 이미지 생성 분야에서 생성형 인공지능의 장점과 단점을 드러내며, 미래의 최적화에 참고자료를 제공하였다. 본 연구는 디자이너가 디자인 과정에서 주도적 위치를 유지하며, 인공지능을 보조 도구로 활용하여 창의성을 확장해야 한다고 주장한다. 인간-기계 협업 디자인이 실현 가능성과 유효성을 가지고 있으며, 디자이너가 생성형 인공지능을 더 잘 활용할 수 있도록 구체적인 제안을 제시하였다.

This study focuses on the collaborative design between designers and generative artificial intelligence, aiming to create images that align with design intentions. In the theoretical background section, the theoretical foundations of the Fourth Industrial Revolution and the development of artificial intelligence are explored. The working principles and functional characteristics of ChatGPT and Midjourney are analyzed in detail, revealing that the reasonable combination of these two tools can significantly enhance the alignment and creativity of image design. From a cognitive perspective, the design thinking process of designers is analyzed, and the design thinking process of this study is summarized, providing experimental guidance for this research. Experimental results indicate that under the guidance of clear design intention prompts, the images generated by ChatGPT and Midjourney align with the designers' expectations. This reveals the advantages and limitations of generative artificial intelligence in image generation, offering references for future optimization. This study suggests that designers should maintain a leading role in the design process, using artificial intelligence as an auxiliary tool to expand creativity rather than replace human creativity. Human-machine collaborative design is feasible and effective, and this study also provides specific recommendations for designers to better utilize generative artificial intelligence.

GPT-3.5를 대표로 하여 초거대 언어모델이 개발되고 있으며, 이는 다양한 분야에서 인공지능 활용 기술의 빠른 발전을 촉진하고 있다. 교육 분야에서도 인공지능을 활용하거나, 인공지능과 교육을 융합하여 적용하려는 노력이 지속되고 있다. 초등학교 교실에서는 기초심리 상담에 대한 요구가 증가하고 있지만, 담임교사가 상담에만 전념할 수 없다는 문제가 있다. 본 연구는 이러한 문제를 해결하기 위해 교사를 도와 기초상담을 할 수 있는 챗봇을 개발 하였다. LLM의 학습모델인 RLHF를 적용하여, '학급질문사전 만들기'를 통해 학생들이 생각하는 인공지능 챗봇에 게 하고 싶은 질문을 수집하고, 프롬프트에 적용하여 제로샷으로 1차 챗봇앱을 개발하고, 학생들이 직접 사용하게 하여, 학생의 질문과 인공지능의 답변을 기록하여 담임교사의 심층상담을 위한 기초자료로 제공하였다. 챗봇을 사용한 학생들에게 설문조사를 실시하여, 그 결과를 프롬프트에 적용하고, 기초 상담자료를 퓨샷으로 2차 챗봇앱을 고도화하였다.

Large language models like GPT-3.5 are being developed, accelerating rapid advancements in various fields of artificial intelligence applications. Efforts to integrate AI into education are also ongoing. Elementary school classrooms are experiencing an increasing demand for basic psychological counseling, but it is challenging for teachers to fully focus on counseling. To address this issue, this study developed a chatbot to assist teachers in basic counseling. Utilizing RLHF, a learning model of LLM, we created a 'Classroom Question Dictionary' by collecting questions that students would like to ask the AI chatbot. We then applied these questions to the prompt to develop a zero-shot first-phase chatbot application. Students used the chatbot, and their questions and the AI's responses were recorded to serve as foundational data for teachers' in-depth counseling. A survey was conducted among the students who used the chatbot, and the results were applied to the prompt to further refine the basic counseling data for the second-phase chatbot application

Purpose: This study aimed to examine the experiences of nursing students developing and implementing generative artificial intelligence (GenAI)-based oncology emergency nursing simulation scenarios. Method: This qualitative content analysis included 23 senior nursing students in the Republic of Korea. Participants worked in teams to develop oncology emergency nursing simulation scenarios using ChatGPT, participated in simulation practice, and completed reflective journals after the program. Data were analyzed using qualitative content analysis. Results: Four themes and twelve categories were identified: “AI-Based Scaffolding for Structuring Clinical Situations,” “Verification of AI-Generated Information and Knowledge Reconstruction,” “Enhancement of Clinical Reasoning and Collaborative Problem-Solving Competencies,” and “Professional Identity Formation and Preparation for Future Practice.” Participants described GenAI as a cognitive scaffolding tool that helped structure complex clinical situations, identify knowledge gaps, critically evaluate AI-generated information, and integrate fragmented knowledge. The program also enhanced clinical reasoning, SBAR-based communication, and collaborative problem-solving. Conclusion: GenAI-based oncology emergency nursing scenario development and simulation may be an effective educational strategy for promoting self-directed learning, critical thinking, clinical reasoning, and practice readiness among nursing students. GenAI can serve as an educational scaffolding tool that supports active knowledge construction and reflective thinking rather than replacing clinical reasoning.

최근 생성 모델이 발전함에 따라 생성 모델을 위협하는 연구도 활발히 진행되고 있다. 본 논문은 Text-to-Image 모델에 대한 멤버십 추론 공격을 위한 새로운 제안 방법을 소개한다. 기존의 Text-to-Image 모델에 대한 멤버십 추론 공격은 쿼리 이미지의 caption으로 단일 이미지를 생성하여 멤버십을 추론하였다. 반면, 본 논문은 Textual Inversion을 통해 쿼리 이미지에 personalization된 임베딩을 사용하고, Adversarial Prompt 생성 방법으로 여러 장의 이미지를 효과적으로 생성하는 멤버십 추론 공격을 제안한다. 또한, Text-to-Image 모델 중 주목받고 있는 Stable Diffusion 모델에 대한 멤버십 추론 공격을 최초로 진행하였으며, 최대 1.00의 Accuracy를 달성한다.

In recent years, as generative models have developed, research that threatens them has also been actively conducted. We propose a new membership inference attack against text-to-image model. Existing membership inference attacks on Text-to-Image models produced a single image as captions of query images. On the other hand, this paper uses personalized embedding in query images through Textual Inversion. And we propose a membership inference attack that effectively generates multiple images as a method of generating Adversarial Prompt. In addition, the membership inference attack is tested for the first time on the Stable Diffusion model, which is attracting attention among the Text-to-Image models, and achieve an accuracy of up to 1.00.

Autonomous AI agents powered by large language models (LLMs) with structured function-calling interfaces have greatly expanded capabilities for real-time data retrieval, computation, and multi-step orchestration. However, the rapid growth of plugins, connectors, and inter-agent protocols has outpaced security practices, leading to brittle integrations ? plugin APIs and protocol adapters that rely on ad-hoc authentication, inconsistent schemas, and weak validation ? making them vulnerable to failures and exploitation. This survey introduces a unified end-to-end threat model for LLM-agent ecosystems, spanning host-to-tool and agent-to-agent communications, and catalogs over thirty attack techniques across Input Manipulation, Model Compromise, System and Privacy Attacks, and Protocol Vulnerabilities. For each category, we provide a formal mathematical formulation of the underlying threat model, defining attacker capabilities, objectives, and affected layers to enable systematic analysis. Representative examples include Prompt-to-SQL (P2SQL) injections and the Toxic Agent Flow exploit in GitHub’s MCP server. For each category, we assess feasibility, review defenses, and outline mitigation strategies such as dynamic trust management, cryptographic provenance tracking, and sandboxed agentic interfaces. The framework was validated through expert review and cross-mapping with real-world incidents and public vulnerability repositories (e.g., CVE, NIST NVD) to ensure practical relevance. Compared to prior surveys, this work provides the first integrated taxonomy bridging input-level exploits and protocol-layer vulnerabilities in LLM-agent ecosystems while introducing formal system definitions for each threat class. Ultimately, it offers actionable insights for securing next-generation AI agents through layered defense and continuous verification. Our work provides a comprehensive reference to guide the design of secure and resilient LLM-agent workflows.

Hate speech ought to be identified and minimized to diminish the undesirable influx of toxic content on the Internet, particularly in the fields of sensitive matters, such as the climate change debate. This article analyzes hate speech detection using climate change-related information and focuses on tweet classification and the identification of the hate target via prompt engineering with large language models (LLMs). Specifically, we compare three instruction-tuned LLMs, TinyLlama, Flan-T5, and Gemma, and evaluate them in contrast to a standard transformer baseline, bidirectional encoder representations from transformers (BERT). The aspects of the experimental framework include prompt design in a systematic way, reproducible implementation environments, and performance measures such as accuracy, precision, recall, and F1-score. The results suggest that prompt-based LLMs for zero-shot inference of hate speech do not require task-specific training. However, the finetuning BERT baseline is more precise and achieves a high F1-score, suggesting the importance of classification reliability. The findings indicate the pros and cons of prompt engineering, which is feasible but highlights the need for prompt design and model choice to achieve reliable hate speech detectors in climate-related speech.

대규모 언어 모델(LLM)의 활용이 증가하며 Prompt Injection 기법을 악용한 Jailbreak 위협이 대두되고 있다. 그러나, 관련 연구는 영어 중심으로 한국어 특화 연구는 미진하다. 이에 본 연구는 LLM의 언어적 취약점을 탐구하고자, 영어, 한국어 평어 체 그리고 한국어 경어체로 작성된 공격의 성공률(ASR)을 비교 분석하였다. 범용 및 한국어 특화 LLM을 대상으로 Roleplay, Parameter 조작, MAC GCG 공격을 수행하였다. 실험 결과, 세 공격 유형과 세 모델 전반의 평균 ASR은 영어와 한국어 평어 체가 각각 약 1.5%, 한국어 경어체가 약 2.0% 수준이었으며, 특히 Roleplay 공격에서는 영어-Llama 조합이 19.23%, 한국어 경 어체-kanana 조합이 10.26%까지 상승하는 등 특정 언어-모델 조합에서 두 자릿수 취약성이 관찰되었다. 이러한 결과는 범용 LLM의 다국어 환경에서의 보안 취약성을 정량적으로 보여준다. 나아가, 경어체 사용이 LLM의 안전 기준 정렬을 우회하는 새로운 취약점으로 작용할 수 있음을 시사한다. 본 연구는 LLM 보안 분석에 사회언어학적 요소를 도입함으로써 다국어 LLM 의 안전성 검증 기준을 심화하고, 한국어 모델의 보안 강화를 위한 실증적 근거를 제공하는 데 기여할 것이다.

The use of large language models (LLMs) is rapidly increasing, and prompt injection and jailbreak attacks are emerging as critical security threats. However, existing studies are predominantly English-centric and research specialized for Korean remains limited. To investigate linguistic vulnerabilities of LLMs, this study compares the attack success rate (ASR) of prompts written in English, plain Korean and honorific Korean. We conduct roleplay, parameter-manipulation, and MAC GCG attacks against both general-purpose and Korean-specialized LLMs. Experimental results show that, across the three attack types and three models, the average ASR for English and plain Korean is approximately 1.5%, while honorific Korean reaches about 2.0%. In particular, under roleplay attacks, the English–Llama combination records an ASR of 19.23%, and the honorific-Korean–kanana combination reaches 10.26%, revealing double-digit vulnerabilities for specific language–model pairs. These findings quantitatively de‘monstrate the security weaknesses of general-purpose LLMs in multilingual settings and suggest that honorific Korean can function as a new vector for bypassing safety alignment. By introducing sociolinguistic factors into LLM security analysis, this study refines safety evaluation criteria for multilingual LLMs and provides empirical evidence to strengthen the security of Korean-language models.

본 연구는 1984년부터 2020년까지 국내·외 프롬프트(Prompts for Restructuring Oral Muscular Phonetic Targets, PROMPT) 관련 연구를 분석하여 우리말에 적합한 프롬프트 프로그램 개발에 필요한 기초자료를 마련하고자 한다. 연구 대상은 1984년부터 2020년까지의 프롬프트 치료기법과 관련된 국외 연구 27편과 국내 연구 1편을 분석하 였다. 분석한 결과, 프롬프트 연구가 처음 시작된 1984년부터 지속적으로 연구가 이루어지고 있으며, 연구방법은 중재 연구가 16편으로, 말소리장애가 가장 많았으며 대상 연령은 3-5세로 유아기를 대상으로 가장 많이 실시되었다. 중재프 로그램은 16회기가 가장 많았으며, 중재프로그램 활동은 무발화 자폐스펙트럼장애 대상을 제외하고 운동구어계층 (Motor Speech Hierarchy, MSH)을 토대로 이루어졌다. 종속변인을 분석한 결과, ‘구어산출‘이 가장 많았으며,’말 운동 제어‘,’조음‘,’구어명료도‘ 순으로 높게 나타났다. 지금까지의 연구를 종합해보면 국외에서는 구어산출에 직접적인 도움을 주는 운동 구어 치료법인 프롬프트가 효과적으로 쓰이고 있다. 그러나 국내에서는 현재까지 우리말에 알맞은 프롬프트 프로그램 개발 및 연구가 부족한 실정이다. 그러므로 본 연구를 통해서 국내 말·언어장애 아동에게 프롬프트를 적용하여 구어산출 및 조음에 도움을 줄 수 있는 우리말 프롬프트 프로그램 개발이 필요하다는 것을 시사한다고 볼 수 있다.

This study analyzed 28 domestic and foreign studies related Prompts for Restructuring Oral Muscular Phonetic Targets treatment techniques from 1984 to 2020 to prepare basic data for the development of PROMPT intervention programs and examination tools. According to the analysis, continuous research has been conducted since 1984 when the prompt study was first started, and the method of research was 16 intervention studies, with the highest number of speech disorders, and the target age being 3 to 5 years old, the most frequently conducted for infancy. The treatment was the most frequent in the 16th sessions, and the activities were based on the Motor Speech Hierarchy(MSH), except for the subjects of the non-verbal autism spectrum disorder. According to the analysis of the dependent variables, ‘speech production’ was the most common, followed by ‘speech motor control’, ‘articulation’, and ‘speech intelligibility’ were highest. Combined with all these studies, it suggests that PROMPT, which are directly useful for exercise spoken word production, are effectively being used outside the country and that it is necessary to develop a PROMPT program that can be applied domestically, in Korea.

This paper proposes a prompt–optimization framework for generating style–consistent game images using Stable Diffusion XL. Given a reference game item image, the system first extracts an initial prompt using a vision–language captioner and a domain–specific prompt bank. The extracted prompt is converted into a list of noun-like elements, and a genetic algorithm searches for compact combinations of these elements under dual gating based on SSIM and CLIP scores. The best combinations are treated as a “style template” that can reproduce the reference image with high structural and semantic similarity. We then investigate whether this template can be reused when the main object is changed while preserving the original visual style. Experiments on fantasy-style item images show that the framework reconstructs reference images using only 8–9 automatically discovered prompt elements, and that changing the main object token together with associated detail elements yields image sets that share a consistent visual style. In contrast, naïvely replacing only the main object token often produces visually ambiguous or stylistically inconsistent images. These results demonstrate that combining automatic prompt extraction from images with evolutionary optimization provides a concrete example of style–preserving prompt design for game item image generation.

This study investigates the impact of different prompt design strategies on the performance of GPT-4 in assessing undergraduate reports within an English as a Foreign Language (EFL) context. As Large Language Models (LLMs) increasingly integrate into educational assessment, understanding how prompt engineering affects grading accuracy and alignment with human judgment is crucial. Three prompt design methods—TELeR Taxonomy, Six strategies framework, and Zero-shot prompting— were applied to evaluate 40 student business reports. The assessments were compared against human teacher scores using quantitative statistical analysis, specifically box-and-whisker plots and inter-rater agreement between teacher ratings and AI-generated scores using Cohen’s κ and quadratic weighted κ. Findings indicate that the choice of prompt design significantly influences the alignment between AI and human grading. The Six strategies method demonstrated strong alignment with teacher assessments, particularly in evaluating coherence and cohesion. The Zero-shot prompt demonstrated the strongest alignment with teacher ratings, yielding the most similar score distribution and the highest inter-rater agreement (quadratic weighted κ = 0.65). The results suggest that although this strategy most effectively approximates human scoring in AI-assisted assessment, human oversight remains essential for capturing the subjective and nuanced aspects of student writing.

본 연구는 PROMPT중재프로그램이 자폐스펙트럼장애 아동의 조음명료도에 미치는 효과를 알아보는 것이 다. 본 연구에서는 자폐스펙트럼장애로 진단되었으며 D지역에 거주하는 만6세 아동 1명을 대상으로 8주 동안 중재를 실시하였다. 연구 결과, 사전·사후검사에서 자음정확도와 모음정확도가 각 39.5%와 30%의 높 은 향상을 보였다. 또한 회기별로 시각정확도와 청각정확도의 정반응을 분석한 결과 회기가 진행될수록 정반응률이 향상되었다. 이러한 연구 결과에 따르면 본 연구는 국내 최초로 자폐스펙트럼장애 아동에게 PROMPT(Prompts for Restructuring Oral Muscular Phonetic Targets) 중재프로그램을 적용하여 조음명료도 의 긍정적인 향상을 보였다는 점에서 의미가 있다고 볼 수 있다. 그러므로 지속적인 연구를 통해 국내의 말 운동 제어를 평가할 수 있는 검사 도구에 대한 연구 및 평가도구 개발, 또한 국내 아동의 발달과정을 바탕으로 활용 가능한 프로그램 개발이 필요하다고 사료된다.

Small Language Models are competitive without large-scale infrastructure, their performance is highly contingent on prompt design. This study analyzes the sensitivity of BitNet b1.58-2B-4T to label exposure and fewshot exemplar composition on a 36-class medical query classification task. We generated 504 items consisting of 6 direct and 8 indirect questions for each disease and after removing cross-exemplar leakage the final evaluation set contained 494 items. With no parameter updates, 0/1/2/5/10- shot prompting was evaluated using Accuracy. Under the nolabel- exposure setting accuracy increased as more exemplars were provided. However, these gains were accompanied by growing prediction concentration on exemplar labels. In contrast with label-exposure, zero-shot achieved the highest accuracy, while the inclusion of exemplars reduced accuracy and amplified label bias. These results show that the structure of the prompt tends to shift few-shot effects from beneficial to detrimental. This highlights the importance of controlled prompt design and domain-adaptive training to ensure trustworthy performance.

Background: Boron neutron capture therapy (BNCT) is a targeted radiotherapy that selectively destroys cancer cells by delivering compounds enriched in boron-10 (10B) that preferentially accumulate in tumors. When exposed to low-energy neutrons, 10B undergoes a nuclear reaction that produces high-linear energy transfer particles that kill adjacent tumor cells while sparing healthy tissues. Real-time monitoring of boron distribution and the neutron irradiation is critical to optimize treatment efficacy and minimize collateral damage. Materials and Methods: We evaluated a single-photon emission computed tomography (SPECT) configuration for simultaneous monitoring of boron spatial distribution and neutron irradiation patterns. Monte Carlo simulations were performed with the Monte Carlo N-Particle version 6.2 (MCNP6.2) using a male Oak Ridge National Laboratory stylized phantom. The simulations modeled prompt gamma emissions at 0.478 MeV (from 10B neutron capture) and 2.2 MeV (from hydrogen neutron capture) and the detection response of a four-head cadmium zinc telluride (CZT) SPECT system. Results and Discussion: The SPECT system localized regions of boron uptake in the phantom using 0.478 MeV emissions with good spatial localization. Although the system detected 2.2 MeV gamma rays from hydrogen neutron capture, reconstructed images from this energy showed diffuse distributions and did not provide clear directional information about the neutron source. These findings indicate strong performance for boron mapping but limited capability for precise neutron-source directionality using 2.2 MeV gamma emissions in the present configuration. Conclusion: SPECT-based imaging with a CZT detector shows feasibility for real-time boron mapping in BNCT. However, further work on detector geometry, collimation, and reconstruction algorithms is needed to improve sensitivity at clinically relevant boron concentrations and to enhance directional sensitivity for neutron-field assessment.

As generative AI becomes increasingly integrated into various applications, its role in enhancing smart home IoT systems has gained significant attention. This study explores the impact of prompt engineering on the performance of generative AI within smart home IoT environments. By optimizing input prompts, we investigate how AI can more effectively control smart home devices, improving performance metrics such as task execution and system efficiency. Using an experimental method, we compare the outcomes of different prompt designs, focusing on key metrics: generation accuracy, completion time, and task performance. Our findings provide insights into how prompt engineering can maximize the utility of generative AI in smart home settings, contributing to advancements in smart home automation and personalized services. This research adds to the growing body of knowledge on AI-IoT integration, highlighting the potential of prompt optimization in real-world applications.