How artificial intelligence affects academic thinking, writing and learning

Written by Xiawan Dong, Hana Farrell and Michael Hogan.

artificial intelligence (AI) changes rapidly how we learn and develop knowledge and skills. By developing artificial intelligence, more and more students are used in higher education AI, such as ChatGPT to help them complete their duties. But how do students and AI cooperate in academic tasks? What are the main reaction patterns and dynamics that make up learning?

Researchers Deng and others. (2025) He was asked recently: Will Chatgpt promote students learning? They conducted a compatible analysis including 69 studies published from 2022 to 2024. Given the total effects in all studies, they concluded that Chatgpt enhances academic performance, emotional emotional states, and higher thinking files, while also reducing mental effort. but, Weidlich et al. (2025) Several disadvantages were identified.

First, different inappropriate interventions have been combined (for example, studies that use ChatgPT to reduce medical terms, write comments, and help unrestricted articles as equivalent) were dealt with. Second, the control groups were diverse, as they did not compare Chatgpt users with groups that did not receive any help, human lessons, or even video games instead of identical instructions without ChatGPT. Third, most studies have failed to measure actual learning, instead using cooperative performance while using ChatGPT, unreliable self -reports, or tests provided during interventions. It is important, studies have failed to measure learning or keep knowledge in the long run after removing ChatGPT.

Decreased mental effort to find Deng et al. Poor analysis is also regarding. Could the higher degrees come at the expense of decreasing cognitive engagement? Roman (2025) It warns that although artificial intelligence provides opportunities for personal learning by delivering adaptive content and increased access and efficiency in the educational process, excessive dependence on risks in artificial intelligence in corrosion through students’ critical thinking skills through a reduced cognitive effort. A recent study conducted Luther and his colleagues (2024) These main concerns are highlighted.

Thinking, writing and learning

The basic knowledge of students and skills development is closely associated with thinking and writing activities, which support the activities of teaching assets and many classroom activities, homework, and evaluation activities. Luther and his colleagues (2024) analyzed the dynamics of student interaction through a writing task, as they presented a detailed look at how students actually used writing tasks on time. It is decisively worried that the patterns of students ’dependency and sharing at the surface level with artificial intelligence outputs that were observed during” thinking and writing “. In the study, 135 participants have completed the task of writing.

Students wrote 600-1000 words discussed Alcohol Prohibition in public places, focusing on the consumption of risky alcohol. During this task, students can cooperate with ChatGPT and they had 40 minutes to write their article. Participants then completed the questionnaires about their experience in ChatGPT and technology rapprochement. Some students also participated in retroactive effects about the mission. Most importantly, while students were cooperating with Chatgpt, all screen activities were registered using the Camtasia 9 program, including all the claims sent to ChatGPT, all ChatGPT responses, and the final texts produced by students. The researchers then coded all the claims of 131 participants who used Chatgpt.

The main results through the diverse student cooperation Dynamic has been determined. The number of claims used by students diverted from 0 to 25 (4 students who have not used Chatgpt at all). Most claims (95.4 %) served as contents related to content, and the vast majority of these requests for data, facts and information-used mainly Chatgpt as a source of information-but also, to a lesser extent, included arguments related to arguments. Distilately, 40.5 % of claims also included requests for full text, and this pattern of claiming behavior of more copies was associated. Interestingly, participants with a higher technique rapprochement were more likely to order complete texts and use a model when processing ChatGPT. Likewise, the current ChatGPT users were likely to request full texts and individual sections compared to non -users.

Students classified a higher Chatgpt on the efficiency of warmth, expressing high satisfaction in general. They report higher levels than perceptionChatgpt -based confidence compared to influence -based confidence. Students were generally confident in the performance of the trusted and consistent, the accuracy of the task, and the understanding, while its classification was also on its stairs as a collaborator (i.e., friendly, nice, compassionate).

The effects of teachers

The scenario of the state of use in the study conducted by Luther and the colleagues of the common scenarios for thinking and writing supported in higher education. The main issue in the Genai era is the nature of the tools that students can reach. In Luther’s study and colleagues, students’ approaches to dealing with artificial intelligence were associated with familiarity, rapprochement and technology. Unfortunately, high familiarity, rapprochement and confidence in Chatgpt efficiency does little to protect students from excessive dependence on Chatgpt outputs and copy behavior and paste when writing. As a round transformer who was previously trained to train in vast text data, ChatGPT output requires an intersecting from other sources instead of immediate acceptance and establishment.

Students need to maintain academic suspicion and check artificial intelligence responses through interrogation and repetitive verification methods. The researchers continue to raise concerns about the accuracy and reliability of Chatgpt, and the risk of using it to retrieve information in higher education. Teachers need students to be actively reminding of the limits of artificial intelligence. In fact, instead of claiming data, facts and information, ChatGPT can be used more constantly as a cooperative tool to enhance critical thinking.

Using situations can be designed where students actively participate in repetitive exchanges, interrogation and artificial intelligence responses. In these cases, Chatgpt works as a thinking partner that supports both the task and the development of knowledge and basic critical thinking skills. The transformation of students from tool -like use towards cooperative methods is a decisive transition that teachers and politicians must facilitate actively.

In the next blog post, we will explore the completely different Genai state and explain why the activity design is important. The study was conducted before Zhang and his colleagues (2025) Discussion groups of five people included only one member reached ChatGPT. Unlike the study of Luther et al. , These students focused on liquidating and interrogating Chatgpt suggestions actively. The group’s accountability has created the effect of the gatekeeper, as a Chatgpt user became responsible for checking the outputs of artificial intelligence before participating with his colleagues.

This social pressure prevented blind admission and local learners such as the Earths that led the generation of artificial intelligence while improving the results cooperatively, indicating the potential of artificial intelligence to increase human intellectual participation rather than replace them. Ultimately, our point of view is that designing a state of use is the decisive thing that should be focused on in the efforts made to use these new Genai tools to enhance students learning and know and develop skills. This is not an easy task, but by adopting a design mentality we can think through educational Options and make the best possible use of these new tools.

Contact the authors: Xiaoyan dongand Hana Farrelland Michael Hogan