دانلود کتاب پیشرفت‌ها و پیامدهای هوش مصنوعی پیشرفته در کاربردهای دنیای واقعی

Generative artificial intelligence (GenAI) has revolutionised numerous sectors by transforming how we communicate and perform daily tasks. This transformation spans fields such as business, research, education, coding, healthcare, and employment. Users can prompt AI to create human-like text, programming code, images, audio, and video. Unlike traditional conversational AI, which relies on predefined responses, Generative AI produces outputs that surpass its initial programming (Lim et al., 2023), enhancing learning experiences, particularly in education. However, ensuring accuracy, avoiding overreliance, and addressing data privacy are essential for its successful integration (Eke, 2023). Despite various challenges posed by these Generative AI tools, including concerns about the impact of Generative AI on originality and assessment reliability (Yu, 2023; Dwivedi et al., 2023), such tools hold significant potential to assist in various educational endeavours (Ausat et al., 2023), the role of Generative AI remains significant. Despite challenges, these tools offer considerable promise, particularly for generating personalized feedback on assessments (Kaiss et al., 2023; Baidoo-Anu and Ansah, 2023). Constructive assessment feedback is critical to improving student learning, though delivering timely, quality feedback for large cohorts remains challenging (Pardo et al., 2019). As a result, students express significant shortcomings in the quantity and quality of the feedback they are provided with (Boud and Molloy, 2013). These challenges have motivated educators to leverage two other sources of external feedback: student peer feedback or automated feedback systems. In student peer review, students evaluate their peers’ work, present a means to foster collaborative learning, encouraging active engagement and deeper understanding through interaction with peers’ insights (Lerchenfeldt et al., 2019). Moreover, peer review can nurture a sense of accountability, as students assume responsibility for their evaluations and contributions to the learning process. This approach also holds the potential to alleviate instructors’ workload, as students take on a more participatory role in assessment. However, unhelpful feedback, as well as biases and inconsistencies in feedback, provided by peers must be acknowledged as its limitations (Lerchenfeldt et al., 2019). Traditional automated feedback systems generate feedback based on predefined set of rules for feedback delivery. The source of this rules is either domain experts (expert-driven), student data (data-driven) or both (hybrid) (Deeva et al., 2021). Available automated feedback systems (such as BEETLE II, ActiveMath, SQL-Tutor, and others) offer several advantages over instructor feedback such as speed and scalability (Deeva et al., 2021). However, they have limitations such as low levels of learner control and adaptiveness to student characteristics, as well as focusing more on feedback automation to address teacher’s workload, rather than mainly addressing the quality of the feedback given to the students (Deeva et al., 2021). With the increasing use of Generative AI in assessment evaluation and generating feedbacks (Teubner et al., 2023), the question arises as to whether it can be effectively used to generate automated feedback and how the usefulness of AI-generated feedback is compared to other types of feedback. Although some studies (Kochmar et al., 2020; Fui-Hoon Nah et al., 2023) reported the utility of intelligent AI tools for providing automated feedback to students and its impact on learning outcomes, studies on the use of LLM-based for providing student feedback are scarce. To gain an in-depth analysis of the effectiveness of feedback generated by LLM-based AI, it is imperative to assess the usefulness of AIgenerated feedback for delivering constructive comment to students.