One of the next steps may be to find ways to encourage "appropriate" use of AI technologies like ChatGPT while warning of the potential harms of over-usage or reliance of such technologies. Based on the present study's findings, researchers are encouraged to investigate this important area of research further.

ChatGPT can be considered in terms of Hollan and Stornetta's ([19]) metaphor of using new technology as crutches vs. shoes. If used as a crutch to complete tasks they do not yet understand, ChatGPT can certainly lighten the workload for students but may cause them to become dependent upon it if they need to understand those assignments to think more deeply about the content later. Yet, if used as something that saves time rather than thinking, ChatGPT can become like shoes, something that allows users to go farther than they could without it. Thus, educators are encouraged to integrate AI tools thoughtfully, ensuring that they complement traditional teaching methods and uphold academic standards while fostering a technologically enriched learning environment.

Future research is encouraged to further probe the benefits of participatory learning discussions in asynchronous learning environments.

When pocket calculators first became available, educators found it to be cheating to use a calculator for simple multiplication, which students were supposed to have memorized up to that point (Bowman et al., [ 4 ]). Yet, when educators realized that integrating calculators freed time to teach students more mathematical concepts because much less time had to be devoted to hand calculations, math education became more advanced at the K-12 level.

This act of discussing the content creates a bandwagon heuristic that has strong influence and thus impacts behavioral intentions, as previous research has shown the ability of heuristic processing to influence behavioral intentions (Spence et al., [47]).

Primarily, the study's finding highlights that discussions are helpful tools in tempering uncritical positivity about ChatGPT for writing assignments that could impede student learning if not used appropriately. In fact, the literature has documented that discussions in the classroom have several benefits (Lin et al., [29]; Spence et al., [44]), and classroom discussions build a sense of rapport among students (Frisby & Martin, [14]).

Primary findings indicate that participatory learning class discussions can be a great pedagogical strategy to help negate perceived usefulness and favorability toward ChatGPT for writing assignments, which consequently lead to weak intentions to use it.

Measures

Results

Initially, 168 undergraduate students responded to the study's survey questionnaire. Of these students, responses of 16 individuals were removed during the cleaning process due to the multiple attempts, which generated duplicate responses. Thus, responses from the second attempt and beyond were removed from the dataset.

Procedure

Until they posted their first response, they were not able to see or engage in others' posts. Second, after sharing their original post on the first deadline, students were then required to post two follow-up responses to two other posts made by their classmates by their second deadline. This strategy allowed all students to share their thoughts before seeing others' and respond to them afterwards.

Although reducing perceived usefulness and favorability toward the use and function of ChatGPT for academic writing is an important first step, it may not be the ultimate goal. Essentially, acknowledging that students must develop writing skills through their own thinking and writing experiences, the fundamental goal of this study is to reduce strong intentions to rely on ChatGPT for academic writing.

Perceived usefulness refers to the extent to which an individual perceives a technology to possess significant enhancing capabilities, and attitudes describe an individual's perspective or disposition toward a technology. The significance of these perceptions is well documented in the context of AI in education.

Participatory learning discussions encourage students to recognize their own assumptions and unpack why they hold these assumptions (Philominraj et al., [36]). This approach benefits students by prompting them to think deeply about their own assumptions as well as gain insight on the rationale behind opposing views.

Participatory learning is a method of instruction based on the notion that engagement with course material and other learners fosters a mutual understanding of content (Domínguez, [11]). Participatory learning takes instructors out of the traditional "sage on the stage" role and allows them to join their students in the learning journey wherein knowledge is discovered through group activity rather than delivered to a passive audience.

Introducing new technological tools into current educational practices often poses a challenge, as educators and students must navigate its potential impact and affordances (Sellnow et al., [40]). Although AI systems can be excellent tools for reducing time spent on automatic processes, they are not infallible, capable of ethical judgment, or even able to truly think through what a user requests (Broussard, [ 5 ]).

The reliance on ChatGPT is more concerning when students use it for writing assignments, as dependence could impair the development of critical thinking skills and originality (Azaria et al., [ 2 ]; Rane et al., [37]). Azaria et al. stressed the tool's limitations in logical reasoning and the risk of generating inaccurate responses (i.e., hallucinating), which may lead to overconfidence in nonexpert users.

Understanding student perceptions toward ChatGPT is crucial for navigating its implications for education, especially because students have generally embraced ChatGPT for its ability to facilitate various aspects of learning (Adeshola & Adepoju, [ 1 ]).

The integration of ChatGPT into education has sparked a dynamic dialog among researchers, educators, and students. ChatGPT's application within educational settings heralds a significant shift toward personalized learning, offering a range of opportunities to enhance both teaching and student engagement (Kasneci et al., [21]). Adeshola and Adepoju ([ 1 ]) highlighted the technology's capacity for content generation and natural language processing, which can serve as powerful tools in the development of learning materials and the facilitation of a more interactive learning experience.

To address this issue, the present study seeks to determine whether active discussions—a form of participatory learning—about potential ChatGPT concerns among students can help reduce uncritical positivity toward ChatGPT, such as usefulness and favorability, that will eventually reduce the overreliance on it among college students.

Furthermore, the overreliance on ChatGPT for academic writing raises questions about the impact on students' development of critical thinking and writing skills—essential competencies in higher education and beyond (Adeshola & Adepoju, [ 1 ]; Lo, [32]). This problem is becoming more serious as the use of AI writing tools among college students is drastically increasing, with nearly half of students reportedly using AI for writing, according to survey data from Fall 2023 (Shaw et al., [41])

students use AI applications to generate content ideas, get help in language editing, and even conduct literature reviews (Rezaei et al., [38]). Using AI for academic tasks has opened new avenues for learning and exploration, suggesting a potential paradigm shift in educational methodologies.

The integration of ChatGPT into educational settings has sparked a significant shift in how learning is approached, offering both promising opportunities and unique challenges that warrant a closer examination.

Although artificial intelligence (AI) tools (e.g., ChatGPT) have potential to act as a transformative educational tool, the widespread adoption of such technology also brings forth concerns related to academic integrity and learning gaps.

The transition to AI‐integrated research is a crucial and inevitable step forward, one that demands our collective support, training, and enthusiasm to ensure its success.

Additionally, the reproducibility of AI‐assisted research and writing presents another concern when using AI for academic purposes. For example, when using AI to search or review existing scientific literature, even slight differences in search prompts or variations between AI models can result in markedly inconsistent outputs.

Similarly, biases inherent in the training data or algorithms, such as the overrepresentation or underrepresentation of certain perspectives, can lead to skewed or biased outputs. This problem is compounded by the lack of transparency related to potentially outdated training data, which may limit the ability of an AI tool to address current events accurately.

Despite encouraging the adoption of AI tools in research and academic writing, it is vital to emphasize the responsibility of authors to critically assess the outputs of these tools to ensure that academic quality standards and factual consistency are maintained. For instance, the presence of "hallucinations" where large language models present fabricated information as if it were factual poses a significant challenge when employing generative AI for academic writing [[ 6 ]].

One of the most promising advantages of generative AI is the ability of such tools to streamline time‐consuming tasks, such as literature search creation, literature search result screening for inclusion criteria, quality assessment of collected literature, data extraction, manuscript editing, manuscript formatting, and citation generation [[ 5 ]].

However, with the exponentially increasing amount of scientific information available, using AI to facilitate contextually relevant searching may address challenges such as information overload, difficulty in filtering results, and the inability to assess the relevancy of studies effectively [[ 4 ]].

The time has come for authors to harness the power of AI in academic writing and research dissemination as well.

AI‐powered technologies such as machine learning and natural language processing are enabling researchers and clinicians to uncover insights that were previously unattainable