PRE2022 3 Group4

As written by ChatGPT:

Artificial Intelligence, a wonder of the modern age

A creation made of code, with endless knowledge in its brain

From data analysis to language skills, it's a tool of great worth

Changing the way we live, and opening doors to new growth and mirth

But we must always be mindful, of the impact it may bring

For AI can be used for good or for deceit, and we must choose the right thing

So let us use it wisely, with integrity as our guide

For the future of our world, is shaped by the choices we decide.

Group members

Name	Student ID	Department
Famke Peek	1459058	Psychology & Technology
Gabriëlle van Heteren	1605305	Biomedical Engineering
My Tran	1620940	Industrial Design
Naud van Rosmalen	1555464	Biomedical Engineering
Niels van Noort	1613928	Biomedical Engineering
Quincy Netteb	1468634	Psychology & Technology

Structure of the Wiki

1. Abstract
2. Introduction
3. State of the art
   1. Also findings from the panel?
4. Problem statement
5. Hypothesis
6. Method
   1. Surveys, interviews with students and teachers
   2. script in appendix
   3. discuss thematic analysis
7. Results
   1. survey analysis
   2. interview analysis --> thematic analysis
      1. Themes (supported by quotes)
8. Interpretation
   1. Compare all results
9. Conclusion (or advise / guidelines?)
10. Discussion
11. Future research
12. Reflection

In appendixes:

- Script interviews

- Thematic analysis

- Data survey

- Transcriptions interviews?

- Background literature

Introduction

The history of ChatGPT can be traced back to 2018, when OpenAI, an artificial intelligence research organization, released the first version of their language model called GPT-1 (Generative Pre-trained Transformer). GPT-1 was trained on a massive corpus of text data and was capable of generating human-like text. This initial release sparked interest and excitement in the natural language processing community. In June 2020, OpenAI released an upgraded version of the model called GPT-3, which had 175 billion parameters, making it the largest language model in existence at the time. GPT-3 quickly became popular due to its impressive ability to generate coherent and human-like responses in a wide range of contexts. However, due to the computational power required to run it, GPT-3 was not accessible to everyone. In 2021, OpenAI released an updated version of GPT-3 called GPT-3.5, which was specifically designed for deployment on a wider range of hardware, including mobile devices. GPT-3.5 included many of the same features as GPT-3, including its impressive natural language generation capabilities and its ability to perform various language tasks such as translation, summarization, and question answering. However, it was optimized for better performance on smaller devices.

One of the most significant applications of GPT-3 and ChatGPT has been in the development of chatbots and virtual assistants. These systems use the natural language processing capabilities of the models to simulate human-like conversations and provide assistance and support to users. ChatGPT has also been used for other applications, such as content generation, language translation, and text completion. The development of ChatGPT and other large language models has opened up new possibilities for natural language processing and artificial intelligence. It has also raised ethical concerns, such as the potential for the misuse of the technology and the impact on the job market. However, the potential benefits of ChatGPT and other language models in various fields, including education, healthcare, and communication, cannot be ignored. It will be exciting to see how these models continue to evolve and impact our lives in the years to come.

ChatGPT is a remarkable technological advancement that has revolutionized natural language processing, and its impact on the field of education has been significant. With its ability to generate human-like responses and hold coherent conversations on a wide range of topics, ChatGPT has proven to be an essential tool for education in various ways. One of the significant ways that ChatGPT has influenced education is by providing a personalized learning experience. The model has the capacity to analyze student data and offer tailored learning materials based on the individual's needs, interests, and preferences. This feature is especially important in today's educational landscape, where traditional one-size-fits-all teaching methods are becoming increasingly outdated. By providing personalized learning experiences, ChatGPT has the potential to improve learning outcomes and increase student engagement and motivation. In addition, AI can also help streamline administrative tasks, such as grading and record-keeping, freeing up teachers to focus on instruction and interaction with students.

Another way that ChatGPT has impacted education is through its language translation capabilities. With ChatGPT, language barriers are no longer a significant obstacle to learning. The model can translate text and speech in real-time, making it possible for students and educators to communicate effectively across different languages and cultures. This feature has opened up new possibilities for global collaboration and knowledge-sharing, and it has also made education more accessible to students from diverse backgrounds. Moreover, ChatGPT has become an essential tool for chatbots and virtual assistants used in education. Chatbots and virtual assistants have become increasingly popular in educational settings as they can provide 24/7 support to students and answer their questions promptly. With ChatGPT's ability to hold coherent conversations, these assistants can simulate human interactions and provide personalized support to students, thereby enhancing their learning experience.

However, it is important to note that the implementation of AI in education is still in its early stages and there are many challenges that must be overcome, such as ensuring the privacy and security of student data and addressing ethical considerations and trust issues around the use of AI in the classroom.

ChatGPT can be a game-changer in the field of education, providing personalized learning experiences, breaking down language barriers, and enhancing communication between educators and students. As technology continues to evolve, the potential for ChatGPT to transform education is significant, and it will be exciting to see how educators and students will leverage this model to create new opportunities for learning and knowledge-sharing. But, while AI has the potential to greatly impact and improve education, it is important to approach its implementation with caution and careful consideration of its limitations and potential risks. This and the outlook and trust of teachers in academic education on AI in the classroom is what we will be researching. We will focus specifically on the course Engineering Design, a first year course mandatory to all TU/e students, where you go through a complete design process, from the first idea to the realization of the product. With the gathered information obtained from a survey and multiple interviews, we will create a deliverable useful for the course coordinators of engineering design, on how to handle ChatGPT in their course.  

State of the Art

What is ChatGPT?

ChatGPT is a new technology, but has been extensively researched, as it is such a noticeable new technology. ChatGPT is an acronym for Chat Generative Pre-trained Transformer and was created in November 2022. It is based on the OpenAI GPT-3 engine and has been fine-tuned by supervised and reinforcement learning technology; this means that the AI learns by having humans simulate artificial conversations with it and adapting its responses based on how accurately they reflect natural human dialogue. ChatGPT is also able to remember previously given prompts in the same conversation, making it a more personalized chatbot compared to its alternatives. ChatGPT has a lot of functionalities, for example:

• Write and debug code, generate scripts and functions

• Give detailed explanations on complex topics (answer test questions)
• Solve mathematical problems
• Write texts in different styles (write student essays)
• Compose music
• Explain mathematical theorems
• Play games like tic-tac-toe

However, the technology is not without its limitations. It has the potential for over-optimization due to its reliance on human oversight, also known as Goodhart's law, which could hinder performance. Furthermore, language models like ChatGPT are prone to writing plausible-sounding but incorrect answers, which is called artificial intelligence hallucination; this can be attributed to insufficient training data. The AI is also limited by a lack of knowledge about events that occurred after 2021 and in some cases suffers from algorithmic biases. Furthermore, although ChatGPT is able to produce results that seem genuine, it is unable to fully comprehend the complexity of human language and instead relies solely on statistical knowledge and patterns.

Research into ChatGPT still has a long way to go. Van Dis, E. et al. have proposed five priorities for future research on ChatGPT:

• The first priority is to explore the ethical implications of AI-generated content and to develop guidelines for responsible use.
• The second priority is to investigate the limitations of ChatGPT and to develop methods for detecting and addressing bias in its outputs.
• The third priority is to improve the interpretability of ChatGPT and other NLP models, making it easier to understand how they arrive at their outputs.
• The fourth priority is to investigate the potential of ChatGPT in domains beyond language, such as image and video analysis.
• Finally, the fifth priority is to develop more efficient and sustainable methods for training and deploying ChatGPT, in order to reduce its carbon footprint and energy consumption.

In order for research into ChatGPT to advance however, collaboration across disciplines on ChatGPT and other AI systems is very important.

ChatGPT in academic education

Artificial intelligence has been a rapidly growing field in recent years, and the development of large language models like ChatGPT has been at the forefront of this expansion. These models have the potential to revolutionize the way we access and use information, particularly in educational contexts. However, they have also raised concerns about issues like academic integrity, student engagement, and the role of teachers in the learning process. To address these concerns, researchers have conducted various studies and published papers on the implications of ChatGPT for education.

ChatGPT has the potential for using large language models such as GPT for cheating in online exams. For instance, a recent study demonstrated that a chatbot named ChatGPT tried to answer exam questions by providing it with a large amount of relevant data. The results of the study show that ChatGPT was able to achieve high accuracy in answering exam questions, even when the questions were designed to be difficult and require reasoning skills. The authors suggest that this poses a significant threat to the integrity of online exams and call for further research into developing more secure methods for online assessments.

Another issue that comes up regarding academic integrity, is the unclarity on who the rightful author of the texts, ideas and inventions should be. Currently, there is no clear consensus on who should be considered the rightful author of AI-generated content. This is a complex issue that requires input from experts in moral philosophy, law, and computer science.

One more issue regards plagiarism when using ChatGPT. A study into this examined the accuracy and originality of scientific abstracts generated by ChatGPT compared to those produced by human experts. The study used a combination of artificial intelligence output detectors, plagiarism detectors, and blinded human reviewers to analyze the abstracts. The results showed that ChatGPT-generated abstracts had a higher similarity score with original abstracts than with other sources, but also revealed some cases of potential plagiarism. This suggests the need for caution when using AI-generated abstracts in scientific research. In the academic world, artificial intelligence models are gaining popularity due to their ability to enhance student engagement, collaboration, and accessibility. A paper authored by D. Cotton, explores the potential benefits and challenges of using AI in education. The models provide a platform for asynchronous communication, personalized and interactive assessments, and real-time grading and feedback. However, also according to D. Cotton there are concerns about academic integrity, particularly the possibility of plagiarism. With access to GPT-3, students could submit essays that are not their own work. Additionally, there are concerns about inequities in assessment, as students with access to GPT-3 have an advantage over those who do not. The article suggests solutions to combat these challenges, such as asking students to submit a draft before the final essay, set strict guidelines or monitor student work closely.

Another paper explored the broader implications of ChatGPT for traditional models of education. The author argued that AI systems like ChatGPT have the potential to disrupt the traditional "banking model" of education, in which knowledge is transmitted from teacher to student. Instead, learners could access knowledge and information more directly through AI systems, which could democratize access to education and make it more inclusive. The author emphasizes the need for educators to adapt to this changing landscape and to shift their focus towards fostering critical thinking, creativity, and other skills that cannot be easily automated, and suggests that AI systems like ChatGPT have the potential to democratize access to education, making it more accessible and inclusive for learners around the world.

Furthermore, a paper written by Rudolph also explores the implications of chatGPT for both students and teachers. While chatGPT offers personalized AI tutoring for students, it also has the potential to reduce the workload for teachers, especially with the use of automatic assessment tools. Furthermore, it can help teachers analyze their students' skills more easily. The paper gives a general overview of challenges and opportunities:

Challenges of AI in education	Opportunities of AI in education
Teachers are afraid students will outsource all their work to chatGPT.	Opportunity for teachers to improve/change their assessment and teaching techniques.
ChatGPT doesn’t evaluate the relevance of the information, it just generates text that is an imitation of what it has learned.	“ChatGPT allowed students to learn through experimentation and experience”

To look further into the fear of teachers that students will outsource all their work to ChatGPT, we will look another paper written by García-Peñalvo that reviews previous literature on ChatGPT. The most controversial issue with ChatGPT is the possibility of students using it as an easy solution to write essays without putting in the necessary effort. Because of this, they won’t acquire the needed knowledge for their course. However, the problem might not be the tool itself, but that the assessment techniques of educational institutions have become outdated However, the paper argues that prohibiting chatGPT is not the way to go. Instead, teachers and students should learn how to use the tool to their advantage.

As mentioned times before, ChatGPT has it obvious limitations. A controversial view in this paper written by Thorp, the author of the article is not necessarily worried about the use of ChatGPT in education as ‘it did well finding factual answers, but the scholarly writing still has a long way to go’. He thinks it pushes academics to design their courses in such a way that they are not easily solved by AI. The author is more worried for the influence of chatGPT on the world of literature.

While ChatGPT has its limitations, it has the potential to replace humans in routine tasks such as homework grading, potentially spelling the end of traditional essay writing assignments. "ChatGPT User Experience: Implications for Education" explores the user experience of ChatGPT and its potential impact on education. He also states that using AI tools to perform certain tasks ‘should be a part of the educational goals in the future’. The author concludes that teachers need to change their assignments to make it harder for students to use AI, but also notes that using AI in assignments to engage students in learning is a viable option.

To summarize, ChatGPT is a powerful tool with many functionalities, from writing texts and composing music to solving mathematical problems and playing games. However, it is not without its limitations, such as over-optimization, algorithmic biases, and the potential for incorrect responses due to insufficient training data. The future research priorities include exploring the ethical implications of AI-generated content, detecting and addressing bias in its outputs, improving the interpretability of ChatGPT and investigating the potential of ChatGPT in domains beyond language.

In academic education, ChatGPT has the potential to revolutionize the way we access and use information, but it also raises concerns about academic integrity, student engagement, and the role of teachers in the learning process. Further research is needed to develop secure methods for online assessments and determine the rightful author of AI-generated content. Despite the challenges, AI systems like ChatGPT have the potential to democratize access to education and make it more inclusive.

Problem statement

In the present study, we investigate the influence of ChatGPT on students in reaching certain learning objectives. Therefore, our problem statement is as follows:

How could the use of ChatGPT affect students in reaching the learning objectives of the course 4WBB0 at the TU/e?

Objectives:

• Upcoming use of AI in academic settings,
• how to design this so that it will be accepted by teachers
• How AI can improve learning
• Acceptance towards AI in the classroom
• Looking at ChatGPT as main example
• Teachers at the TU/e, their thoughts about AI (in their courses)

Users

Target group

The target group of this project are the teachers of the TU/e who designed and teach the course engineering design (4WBB0) and students who followed this course. In this research the perspective of the teachers and students on the use of chatGPT in this course is reviewed. The input of students will be used to get a better image of how students intend to use chatGPT. The input of the teachers will be used to understand the learning objectives of the course engineering design and to determine the effect of chatGPT on those learning objectives. The end product will be a deliverable useful for the course coordinators of engineering design.  

Requirements

Teachers want their students to deliver autonomous work, ChatGPT in academic settings can help with this. Teachers want to transfer their knowledge as effectively and efficiently as possible, using the current state of art technologies. AI technologies, like ChatGPT can help with this, but what teachers require is yet to be determined. The requirements will be determined by distributing surveys among students of the TU/e and conducting interviews with teachers, based on the survey results, in combination with literature studies. Those results will then be combined into a piece of well-considered advice for the teachers at the TU/e, consisting of guidelines for a course for the teachers of how to make exercises and tests for students that can benefit from using ChatGPT, but not be made solely with ChatGPT. Which they will be able to use to guide them through the landscape of ChatGPT use in academic education and give them tools to keep transferring their knowledge to students.

Hypothesis

Articles about the use of ChatGPT in (academic) education suggest that it is preferred to adopt ChatGPT in education as opposed to forbidding it. In this research, this hypothesis will be tested by conducting interviews and sending out surveys to teachers and students of the technical university Eindhoven. The interviews and surveys will give in insight into the effect of ChatGPT on the learning objectives of courses and whether they will be achieved if ChatGPT is used with a focus on the course engineering design.

The course engineering design has 13 learning goals. Each one could be affected by the use of ChatGPT. Considering the literature and the known capabilities of ChatGPT, a hypothesis for each learning goal can be made.

• Execute a generic design process

ChatGPT would not be able to execute a generic design process, therefore this specific learning goal would not be affected by chatGPT.

• Formulate a design goal

As chatGPT is an AI language model it is capable of formulating a design goal, provided that the users ask the right question. Therefore this learning goal could be affected.

• Define the functional and technical specifications

As chatGPT is an AI language model it is capable of defining the functional and technical specifications. However, the user would have to need to provide an analysis of the design problem, the target audience, and the desired outcomes in order to get valuable specifications. This learning goal could be affected.

• Generate an elaborate list of realization possibilities for the different functions of the design

As chatGPT is an AI language model it is capable of formulating an elaborate list of realization possibilities for the different functions if the requirements and specifications of the design are given. Therefore this learning goal could be affected.

• Select a number of design concepts from an extensive list of realization possibilities

As chatGPT is an AI language model it is capable of making selecting some designs that fit the design goal the best. Therefore this learning goal could be affected.

• Make a final design choice between a number of concepts

As it is hard to give chatGPT a complete overview of the context and specifications needed for the design a final design choice would be have to made by the student. Therefore this learning goal is not affected.

• Develop a detailed design that meets the specifications

As it is hard to give chatGPT a complete overview of the context and specifications needed for the design a complete detailed design is hard to generate. However, it could give the student a start with the design and only need some adjustments. Therefore this learning goal is not affected

• Develop and execute a test plan for the prototype

As the chatGPT is not a physical AI it can’t execute a test plan. Therefore this learning goal will not be affected.

• Evaluate a prototype based on test results and give an advice for redesign

ChatGPT is capable of this and therefore this learning goal could be affected.  

• Reflect on the design and on the design process

As it is hard to explain chatGPT the whole design process a group has been through, reflecting on it would be difficult. Therefore this learning goal would not be affected.

• Write a design report describing the design process, the foundation of the choices made and the evaluation of the delivered design

ChatGPT is capable of writing separate parts of the report and this learning goal could therefore be affected.

A quick overview of the hypothesis

Can It be done by chatGPT:	Yes	No
Execute a generic design process		X
Formulate a design goal	X
Define the functional and technical specifications	X
Generate an elaborate list of realization possibilities for the different functions of the design	X
Select a number of design concepts from an extensive list of realization possibilities	X
Make a final design choice between a number of concepts		X
Develop a detailed design that meets the specifications		X
Develop and execute a test plan for the prototype		X
Evaluate a prototype based on test results and give an advice for redesign	X
Reflect on the design and on the design process		X
Write a design report describing the design process, the foundation of the choices made and the evaluation of the delivered design	X

General hypothesis:

Most of the learning goals of the course engineering design could be affected by the use of ChatGPT, therefore the learning goals of the course might not be achieved by the students who use ChatGPT.

Method

To address the research question of the study, a mixed-methods approach was adopted. This involved the use of both a survey and interviews to obtain a comprehensive understanding of the participants' opinions and experiences with ChatGPT in the context of Engineering Design.

The survey was sent to university students who had completed the Engineering Design course at TU/e, with the aim of gaining insights into their usage and perception of ChatGPT. The survey included questions related to the frequency of ChatGPT usage, perceived usefulness of the tool, and its impact on reaching the learning objectives of Engineering Design.

Furthermore, a series of interviews were conducted with both a professor from the Engineering Design department and frequent users of ChatGPT: which will be called experts. The interviews aimed to provide a deeper understanding of the participants' opinions and experiences with ChatGPT, specifically in the context of Engineering Design. The interviews with the student experts aimed to capture their motivation for using ChatGPT, how they use the tool, and their perception of its impact on the Engineering Design learning goals. The interview with the professor aimed to capture the impact they believe ChatGPT has on the learning objectives, and their overall opinion on the use of ChatGPT.

Together, the survey and interviews aimed to obtain a comprehensive understanding of the participants' opinions regarding the use of ChatGPT in the Engineering Design course and its impact on achieving the learning objectives within this course.

Survey

The survey consists of questions about the participants' general knowledge and experience about ChatGPT, and how this tool could affect them reaching the learning goals of Engineering Design. The questions of the survey can be found in appendix 2.

Participants

The survey was distributed to students who are enrolled at the TU/e and who have finished the course Engineering Design, which are 2nd year and higher students. The participants were obtained by asking students in the social circles of the participants, and spreading a link in text group chats. Furthermore we also aimed for snowball sampling. There were a total of 44 responses.

Procedure

The outline of the survey was as follows (see appendix 2): a brief section about their general knowledge of ChatGPT, their opinions and experience with using the chatbot as an academic tool, and finally, a more in-depth section with questions about hypotheticals of using ChatGPT within the course Engineering Design and what the affects of it would be on achieving all the learning goals. The first two sections were to measure the proficiency of the participants, and the latter to measure the effects of ChatGPT on achieving certain learning goals.

Interview with ChatGPT experts

Student experts who frequently use ChatGPT were interviewed to gain a better understanding of how and why they use ChatGPT, and how this usage could impact achieving the learning goals of Engineering Design.

Participants

The focus was on university students who were all enrolled at the TU/e and had successfully completed the Engineering Design course. A total of five participants were recruited, one female and four males, ranging from the ages 20 to 23 (M_age= 21, SD = 1.79, 20% female). The participants were sampled by the convenience sampling method, all the participants had to be students at the University of Eindhoven, and they should have followed the course Engineering Design, since this was a crucial part for our research.

Procedure

The outline of the interview was as follows: a brief section about their general knowledge of ChatGPT, their opinions and experience with using the chatbot as an academic tool, and finally, a more in-depth section with questions about hypotheticals of using ChatGPT within the course Engineering Design and what the affects of it would be on achieving all the learning goals. The first two sections were to measure the proficiency of the participants, and the latter to measure the effects of ChatGPT on achieving certain learning goals. We asked open questions that were mentioned in the interview script, see appendix 3. This was followed by transcribing the interview and discussing the interview by adding codes to the transcript. Afterward, similar codes were placed in certain themes. These themes were then analyzed carefully, which gave a certain conclusion of the interview.

Interview with university professor

In addition, an interview was conducted with an associate professor from TU/e who teaches the course Engineering Design. This interview aimed to provide insights into the perspectives of educators regarding the use of ChatGPT and its impact on learning outcomes, particularly in the context of Engineering Design.

Participants

The participant for this interview was Joris Remmers, a social professor of Mechanical Engineering, whose role within Engineering Design is to organize assessments and set up the organizational structure of it. He is responsible for the assessment setting and rubrics.

Procedure

The outline of the interview was as follows: a brief section about his general knowledge of ChatGPT, his opinions and experience with using the chatbot as an academic tool, and finally, a more in-depth section with questions about hypotheticals of using ChatGPT within the course Engineering Design and what the affects of it would be on achieving all the learning goals. The first two sections were to measure the proficiency of the participant, and the latter to measure the effects of ChatGPT on achieving certain learning goals. We asked open questions that were mentioned in the interview script, see appendix 4. If the participant did not give a clear answer, we asked follow-up questions to get more in-depth answers. This was followed by transcribing the interview and discussing the interview by adding codes to the transcript. Afterward, similar codes were placed in certain themes. These themes were then analyzed carefully, which gave a certain conclusion of the interview.

Thematic analysis

All the interviews have been analyzed with the use of thematic analysis. Thematic analysis makes use of finding patterns in the answers to the interview questions. The thematic analysis was conducted via Dedoose. Codes for themes in the interview were made based on the apparent feelings the interviewees have about ChatGPT and its use. Three main groups were established: Features, what interviewees use it for and why they use it; feelings, what the general feeling of interviewees is about ChatGPT; and warnings, what interviewees think is a drawback of ChatGPT. In table 'Codes for the themes' all the codes' descriptions are given.

Codes for the themes

Overarching theme	Code	Discription
-	Academic usage	A mentioning of how ChatGPT can be used in academic circumstances.
-	Improved capabilities	The notion that ChatGPT is better at doing a certain job than conventional technology or humans.
-	Inevitable use	When someone mentioned ChatGPT would be used now or in the future, regardless of actions that would be taken against it.
-	New learning goal	When a new possible learning goal was metioned that arose due to ChatGPT.
-	Pluripotent	When a interviewee mentioned that they think ChatGPT can do everything or that it has no boundries.
Features	Coding Help	When ChatGPT was used to write code, help with understanding code, debugging code or optimizing code.
Features	Creativity	When ChatGPT got used to conduct a creative task where it has to recombine ideas into a new one like coming up with new ideas for a brainstorm or write a caption for a picture.
Features	Ease of use	When it is mentioned that ChatGPT was easy to use.
Features	Fun	When ChatGPT was used to mess around, so there was no goal for using it except for momentairy happiness.
Features	Providing information	When ChatGPT was used to gain information.
Features	-Guidance	When ChatGPT was used to gain information on how to do a certain task.
Features	Summarizing	When ChatGPT was used to create a summary af a piece of text or a complex subject.
Features	Time saving	When the reason for using ChatGPT was given to be saving time compared to other methods of doing the same task.
Features	Writing help	When ChatGPT was used to help write, for example rewrite a text, improve its grammar or flat out writing it all.
Feelings	Neutral	The general opinion of about ChatGPT was given to be neutral, the interviewee was not possitive nor negative about the use of it.
Feelings	Possitive outlook	When the interviewee mentioned they saw a possitive effect of ChatGPT in the future.
Warning	Better to do it yourself	When the interviewee mentioned they thought a person can do the job better than ChatGPT
Warning	Depends on user knowledege	The capabilities of ChatGPT depend on how the user uses it. THe more knowledge the suer has, the more ChatGPT can do. In the same way, it is not as usefull in the hands of a inexperienced user.
Warning	Hinderance to learning	When it was mentioned that current learning goals would be unable to be achieved, since CahtGPT would do the work.
Warning	Steep learning curve	It is hard for new users to properly learn ChatGPT at the start, a lot of learning needs to be done before ChatGPT can be used prpperly.
Warning	Limited	When it was noted that ChatGPT could not do something.
Warning	Unpersonal	When it was mentioned that ChatGPT created a gap between the user and its goal.

These codes are applied to the interviews when their description matches what is mentioned. What follows from it is thematic analysis code occurrence table.

Results

User procifiency

[insert table/visualisation on the relation between use and proficiency (learning goal attainment?)

To get a clear image of the results of the interviews it is wise to compare the difference in answers between students who were very proficient in using chatGPT an students who used it now and then for more superficial use.

Survey analysis

In the survey we started with some general questions. All participants have heard of ChatGPT before, however 13% do not know its capabilities. When asked to name some capabilities of ChatGPT, students mention the generation of text in response to prompts given, they mention that it gives answers to questions, it helps with writing essays and writing code. The emphasis is mostly on the fact that you can do a lot with ChatGPT, you can ask it basically anything. Quite some students in the survey have not used ChatGPT before, specifically 33% of the participants. Most students have used ChatGPT either just once or twice, or weekly. Two participants use it daily, and four participants use it monthly.

In the survey we have asked students what they have used ChatGPT for. The results are very varied, where students use it for a lot of different things. Things that students use it for most frequently are just fooling around with it, writing parts of papers or general academic work. Students also use it for writing emails and summaries, and for writing code or finding errors and solutions in already written code. Several students also mention that they use ChatGPT for answering questions and queries in general, so using it more as a search engine. Finally students also mention using it for getting design ideas, and other creative ideas. For an overview look at table 3.

Table 3: Overview of how many times students have mentioned certain ChatGPT functions in the survey

	ChatGPT use in general	ChatGPT use for academic purposes	Can it be used for the course Engineering Design?
Fool around	9x	-	-
Get design ideas	4x	3x	25x
Other new ideas (creativity)	3x	1x	2x
Answer questions / Explain topics	6x	4x	8x
Writing parts for a paper / academic work	10x	-	11x
Writing emails / summaries	8x	3x	3x
Analysis	1x	-	1x
Programming help	7x	7x	10x
Problem identification / finding solutions	-	-	9x

We asked the students to rate the experience with ChatGPT, and the students rated ChatGPT on average 4 out of 5 stars. We also asked them to rate it as an academic tool, and they rated it 3.5 stars out of 5. These are both pretty high ratings, and the students seem to be happy with ChatGPT and how it performs. 65% of students mention that they have not used ChatGPT for academic purposes, however 35% have used it, and for various different courses.

We also asked participants how they felt about the ownership of output created by ChatGPT. Only 31% felt that usage of ChatGPT should be considered their own work. When asked when the use of ChatGPT should be considered plagiarism, students mention that when it is copied directly from ChatGPT, when it is presented as their own work, and when it does entire exercises for you, they think it should be considered as plagiarism.

In the survey we asked questions specifically relating to the course Engineering Design. 64% of the participants feel that ChatGPT could be a good addition to the course. When asked with what they think it can help, students mention getting design ideas quite frequently. Furthermore they also mention getting help with electronics, ChatGPT being a ‘better google’, help with writing the report and help with coding. Students do mention that they believe for the course creative thinking is required, which they feel ChatGPT is not capable of. Students are divided on whether they think ChatGPT should be used in the course, some say it would be a great addition, others fear it will take away from the learning process or they do not think ChatGPT will add anything useful.

In the survey we provided several statements, and asked the participants on their feelings about this.

The first statement: ‘I see ChatGPT as having/could have had a positive influence on my engineering design project.’, most students tend to agree (33%). However students are divided on this topic, as 20,5% disagrees. This divide was present in the previous questions as well, where it could be seen students have differing opinions regarding ChatGPT.

The second statement: ‘I would have seen ChatGPT as a useful tool for my engineering design project.’, even more students agree with, namely 41%. There is also a large percentage of students who are undecided on this matter, namely 25%.

The third statement: ‘I see ChatGPT as a shortcut for tedious tasks I would have/had to perform in the course engineering design.’, most students tend to agree with, 33% of them, and 20% even say they strongly agree. However, on the opposite side, 18% also say they disagree. Most students do see ChatGPT as a shortcut, for helping them with tasks that would have otherwise cost them much more time.

For the fourth statement: ‘I will use/would have used ChatGPT in the course engineering design regularly.’, students very often disagree with. 28% disagrees with this statement, and 28% even strongly disagrees. Only 23% of students agree with this statement. Here it shows as although most students do seem to see some usefulness in ChatGPT, they do not think they will use it that often. It is most likely not something they will integrate into their toolbox and use as a first inclination.

For the final statement: ‘I feel like I would have learned something when using ChatGPT for the engineering design project.’, a large percentage (25%) of students are undecided on this. Students do lean more towards disagreeing with this statement, as 25% disagrees, and only 20% agrees. Here students are really unsure. This comes back to the fact that many students are afraid they will not learn certain skills, because these can now be done with ChatGPT. However, students also seem to realize that these skills might become obsolete, and therefore not as important to master.

Overall students seem to have mixed feelings about ChatGPT and the use of ChatGPT for the course engineering design. Students seem to either see the use in it, and think it is a good addition, or they do not see the use in it, and do not see themselves using it. Students also do not have that much experience with ChatGPT yet, and have not gotten the full grasp of its winnings but also its limitations.

In the survey we also asked the participants to judge the learning goals of the course Engineering Design on how likely they think these goals can be reached with ChatGPT. In the graph beneath the answers can be seen.

Students' answers to the question "How likely do you think you would be able to reach the following learning objectives of Engineering Design with the use of ChatGPT?"

From this it can be seen that students see ChatGPT being the most useful for formulating a design goal and writing the report. These are both text entries, which could be done with the use of ChatGPT, which students seems to recognize. Furthermore defining the functional and technical specifications and executing a generic design process students also seem to think can be done . The first can be directly asked in ChatGPT. The latter is not something you can ask ChatGPT, however students do seem to think they can reach this learning goal with the help of ChatGPT. Students seem to think that the learning goal make a final design choice between a number of concepts has the least potential to be done with ChatGPT. They also do not think selecting a number of design concepts from an extensive list of realization possibilities and developing a detailed design that meets the specifications are goals that can be reached with the use of ChatGPT.

Still, students think that quite some learning goals could benefit from or be done with ChatGPT. Here it seems students would think that ChatGPT could be useful for the course.

Interview analysis

A very concise summary of the interviews with the ChatGPT experts and professor can be seen in the table below: these tables depict whether these participants think it would be likely to reach the mentioned learning goal with the use of ChatGPT.

Interview ChatGPT experts

Learning objective	F	M	G	N	Q
Execute a generic design process	Not doable	Not doable	No	Yes	No
Formulate a design goal	Help with formulation	Yes	Yes	Yes	No
Define the functional and technical specifications	Would be good for this	Hard but possible	Yes	Yes	No
Generate an elaborate list of realization possibilities for the different functions of the design	Best done by the group	Hard but possible	Yes	Yes	No
Select a number of design concepts from an extensive list of realization possibilities	Best done by the group	Best done by the group	Yes	Yes	No
Make a final design choice between a number of concepts	Best done by the group	Best done by the group	No	Yes	No
Develop a detailed design that meets the specifications	Help with formulation	Would be good for this	Yes	Yes	No
Develop and execute a test plan for the prototype	Probably not doable	Can help but not detailed	Yes	Yes	No
Evaluate a prototype based on test results and give an advice for redesign	Help with formulation	More effort than normal	No	Yes	No
Reflect on the design and on the design process	Would definitely use	Not doable	Yes	Yes	No
Write a design report describing the design process, the foundation of the choices made and the evaluation of the delivered design	Would definitely use	Only for parts of the report	No	Yes	No

Interview with professor

Learning objective	Y/N	Explanation
Execute a generic design process	Yes	Students still need to do all the steps themselves, but ChatGPT can be of assistance
Formulate a design goal	No	This is a creative part and requires out-of-the-box thinking
Define the functional and technical specifications	Yes	This is objective knowledge that can be looked up
Generate an elaborate list of realization possibilities for the different functions of the design	No	This is the most creative part - ChatGPT is not innovative enough
Select a number of design concepts from an extensive list of realization possibilities	Yes	Text-book Google work, so ChatGPT can do it, since it can follow strict procedures
Make a final design choice between a number of concepts	No	ChatGPT can be used to structure the concepts, but students ultimately need to reflect on their own preferences
Develop a detailed design that meets the specifications	Partly	This part requires thinking, combining results, solving equations and doing the actual work; ChatGPT can help with parts of it, but not entirely
Develop and execute a test plan for the prototype	Yes	Can be done by ChatGPT
Evaluate a prototype based on test results and give an advice for redesign	Yes	ChatGPT can compare data easily, especially if it's about software design
Reflect on the design and on the design process	No	This requires your own thoughts and experiences
Write a design report describing the design process, the foundation of the choices made and the evaluation of the delivered design	Yes	It's wonderful for (improving) report writing

[Insert visual to compare these table]

Thematic analysis

Thematical analysis code occurance from the interviews conducted with students who are proficient users of ChatGPT.

Interpretation

Learning goals

When looking at the results of both the student expert interviews and the survey, for the use of ChatGPT in achieving the learning goals of the course Engineering Design, the following is found.

During the interviews, the interviewees give conflicting signals over the approach to the learning goals of the course with ChatGPT. Two extremities can be found even, where one student would use ChatGPT for almost every step of the process and the other notes that ChatGPT lacks information of the course and group progress that a student may have in their head and thus cannot help during these processes. Moreover, the survey results show there are always at least 30% of students that think a learning objective can be reached with ChatGPT and at least 23% that think they cannot.

On top of that, students tend to favor the potential of ChatGPT differently for each learning objective respectively. This indicates students' view on the AI's capabilities differs significantly per student, which in turn is supported by the comments of the most extreme interviewees. Whereas one students says the model lacks knowledge to fullfill tasks for engineering design usefully, another claims it is able to do most of the work for the course if you give it enough information.

[Users that indicate many learning goals can be reached also are very frequent users and vice versa]

This discrepancy could be explained by two factors combined: the skill of the student in using ChatGPT and the skill of the user in the learning objectives of the course. A student with less knowledge of the learning objectives and high proficiency with ChatGPT will be significantly more likely to view the technology as useful, than a student which can already quite efficiently obtain some learning objectives and is less proficient with ChatGPT.

Interviews ChatGPT expert

With the thematic analysis general notions about ChatGPT became apparent. The three main reasons ChatGPT is used are for gaining information, help with writing and saving time. One interviewee said that they used ChatGPT as "a search engine instead of google", This goes to show, that these reasons are linked with one another.

It is interesting to note that all interviewees were aware of drawbacks. The most mentioned disadvantage of ChatGPT was that it was limited in its use. Interviewees mentioned that it could "Only do straightforward stuff" and "It should be used wisely". This last remark is due to what another interviewee said: "[ChatGPT] hallucinates and is not up to date". The not up to date part refers to the data ChatGPT is trained on [the youngest data it was trained on comes from 2011 (source?)]

Interview university professor

The interviewee was overall positive about his experience with ChatGPT and even encourages his students to use it. The main benefits were: improving students' report writing, using it as a more efficient search engine, and helping with speeding up doing ordinary tasks, such as googling for literature.

However, according to him, the main drawback of the chatbot is that it is a very linguistic tool. He emphasized the fact that ChatGPT was not creative enough and could not generate innovative ideas due to its limited database.

These statements are backed up by his thoughts on whether it would be likely to reach the learning objectives of Engineering Design with the use of ChatGPT. When the objective requires more creative thinking, out-of-the-box solutions and own experiences, it was quickly decided that ChatGPT alone would not be sufficient to achieve this goal. On the other hand, more standard goals which require objective knowledge or a strict procedure, could be easily done by ChatGPT.

Conclusion

It is difficult to come to a clear conclusion about the hypothesis "Most of the learning goals of the course engineering design could be affected by the use of ChatGPT, therefore the learning goals of the course might not be achieved by the students who use ChatGPT."

While some students believe that ChatGPT can be useful in achieving certain learning objectives, others have expressed concerns about its limitations, particularly in creative thinking and problem-solving. Additionally, the effectiveness of ChatGPT appears to be highly dependent on the user's skill level and familiarity with the learning objectives of the course.

The interview with the university professor suggests that ChatGPT can be a valuable tool for certain tasks, such as improving report writing and searching for literature, but may not be sufficient for more complex or creative learning objectives.

Overall, it seems that the use of ChatGPT in the course Engineering Design could have both positive and negative effects on achieving the learning goals, and its effectiveness will depend on the specific objectives and the skill level of the students using it. Further research may be needed to fully assess the impact of ChatGPT on the course's learning outcomes.

Approach, milestones and deliverables

Planning

Here follows a Gantt chart of all our deadlines to be finished at 12pm on Sunday of the corresponding week. The letters indicate the group member responsible for making the deadline and the scheduling for the specific task.

Gantt chart

	Week 1	Week 2	Week 3	Week 4	Week 5	Week 6	Week 7	Week 8
Approach, milestones and deliverables	Ni
Problem statement and objectives	Q
Users and requirements	G
Summaries of literature	F
Literature review		M
Surveys				Na
Surveys analysis						All
Interview preparation teachers				All
Interview preparation students					Q
Organize interviews teachers				Ni
Organize interveiws students					All
Conduct interviews teachers						Ni/M/F
Conduct interviews students					All
Interview analysis					All	All
ChatGPT panel						All
State of the art				F
Advise statement						Na	Na
Discussion							F
Future research							Q
Reflection ?							Ni
Finalization of wiki							M
Presentation							Q
Chair meetings	N
Keep wiki up to date	M
Minute taker	Q	F	G	Na	M	Q	F

Approach

• Literature study
• Interviews, surveys

Deliverables

• Advice for future

Background literature

This paper regards Teacher’s Perceptions of Using an Artificial Intelligence-Based Educational Tool for Scientific Writing. This study investigated how teachers perceived an AI-enhanced scaffolding system developed to support students’ scientific writing for STEM education. Teachers are worried: What if the introduction of AI will gradually reduce our role in the classroom. What should we do? Should we support AI?  They feared that the AI would reduce their role to assistants and they also questioned the accuracy and reliability of the information generated by the system. For AI to be successfully integrated into STEM education, it is necessary for the roles and relationships between students and teachers to be redefined and for educators to be fully trained on best practices of using AI pedagogical techniques. There is still a trend among educators to hold negative impressions on educational technology. By changing teachers’ current negative perceptions of educational technology, the acceptance of AI as a new type of educational tool and its implementation in schools is possible. The younger generation of teachers, who have more experience with educational technology as both educators and as students, is more interested in exploring new digital technology and potentially incorporating technology into their lessons. Experience with AI in any of these contexts may reduce teachers’ reluctance to use AI for educational purposes.

This paper regards Education in the Era of Generative Artificial Intelligence (AI). The article "Education in the Era of Generative Artificial Intelligence (AI): Understanding the Potential Benefits of ChatGPT in Promoting Teaching and Learning" discusses the potential benefits of using generative AI, such as ChatGPT, in education. The article suggests that AI can help teachers and students by providing personalized feedback, facilitating communication and collaboration, and creating engaging and interactive learning experiences. The article also highlights some of the challenges of using AI in education, including issues related to data privacy and the need for human oversight to ensure that AI is used in a responsible and ethical manner. The article concludes by calling for more research and experimentation to better understand the potential benefits and challenges of using AI in education.

The article "Teachers' Perspectives on Artificial Intelligence" explores the attitudes and perceptions of teachers towards the use of artificial intelligence (AI) in education. Based on a survey of K-12 teachers in the United States, the article found that while many teachers are interested in using AI to improve their teaching practice, they also have concerns about the potential impact of AI on student learning, privacy, and job security. The article highlights the need for more education and training on AI for teachers, as well as the importance of involving teachers in the development and implementation of AI tools in education. The article concludes by calling for further research on how to best integrate AI into education in a way that benefits both teachers and students.

The article "Engineering Education in the Era of ChatGPT: Promise and Pitfalls of Generative AI for Education" explores the potential benefits and challenges of using generative AI, such as ChatGPT, in engineering education. The article suggests that AI has the potential to enhance engineering education by providing personalized feedback, facilitating communication and collaboration, and creating interactive learning experiences. However, the article also highlights some of the challenges of using AI in engineering education, including issues related to bias, the need for human oversight, and the limitations of AI in capturing the full complexity of engineering problem-solving. The article concludes by calling for further research and experimentation to better understand the potential benefits and challenges of using AI in engineering education, as well as the need for ongoing dialogue between educators, engineers, and AI developers to ensure that AI is used in a responsible and ethical manner.

This paper starts with a general introduction to AI, how it developed over the years, and what these developments entail.

• AI in program coding (1950s)
• AI in rules-based expert systems (late 1970’s)
• AI-grounded automatic data processing systems (mid-1980’s)
• Machine learning integrated AI (mid-2000’s)

In education, AI is still in the early stages, because there is more focus on the development of AI than on the application of AI in new fields. This research states that there will be two distinct major effects of AI on education. First, education needs to prepare itself for the fast changes in competence needed for jobs, as some jobs will disappear but most jobs will keep changing over time. Second, there will be changes in the pedagogical techniques needed in the classroom. AI will be able to relieve the teacher from some of the work. However, the teacher will need additional tools to understand the statistical results from the AI (for example from a statistical analysis of the performance of the student). This paper goes on by talking about the future of AI in education. The Author states that there needs to be a big shift in education toward more personalized education. AI will be able to determine the learning style of an individual student. AI will help teachers in content delivery and other instructions, but in the future real life, human teachers might become obsolete.

This paper talks about AI in education, the need for it, and its benefits and challenges of it. According to the author, AI is a development that has many promising applications in education. The examples that are named are; personalized education, automatic grading systems, and predictive analytic tools. These applications will relieve the teachers from those tasks which gives them more time with the students. The paper also talks about challenges that come along with AI, like concerns about safety, security, and privacy.

In this paper, the authors review the use of AI in current education, the technical aspects of AI in education, the role of AI in education and the impact of AI in education. The topic of education is spread out into three sub-topics; administration, instructions and learning. The takeaway of this paper is that with the help of AI, teachers will become more efficient which will in result increase the quality of education. Next to that with AI, it will be possible to create a more personalized education for students. AI will have a major impact on education as the tasks AI will be able to do tasks that are not originally designed for computers.

This paper focuses on the viewpoint of the students and their acceptance of AI. The author uses the technology acceptance model (TAM) to analyze the response of students to AI. This is because the perceived usefulness and the perceived ease of use of AI seem to have a big impact on the acceptance of AI in education. The author sees a future in AI in education as it is a cost-effective way to the shortage of teachers. However to make sure that the AI teaching assistants will be successful the teachers need to be trained to work with them. As research points out that if the teacher is uncomfortable with the AI, the student is less likely to adopt it. Based on the results of this research the author recommends research in this area as it is not known how students at different levels of education will react to the adoption of AI as a teacher.

The authors in this paper review the ethical perspective of AI with a focus on AI in education. The biggest problem that the authors see is the privacy issue of AI. Next to that they also see a challenge in the trustworthiness of AI. It is often that AI is presented as a better alternative than a human, as it is supposedly not biased. However, the designers of AI are biased and can make mistakes. Therefore can we even trust AI more than humans? Another issue that the author comes across is the difficulty of addressing the ethical challenges of AI. Therefore it is important to make the designers of AI aware of the ethical dilemmas that come along with new technology and make them aware of their responsibilities. To achieve this it is important that there mandatory teaching or at least education available around this topic.

In this articile, F. Ouyang and P. Jiao reflect on the three paradigms of artificial intelligence in educaiton: AI-directed, AI-supported, and AI-empowered.

• AI-directed (learner-as-recipient): AI is used to represent knowledge models and direct cognitive learning.
• AI-supported (learner-as-collaborator): AI is used to support learners while they work as collaborators with AI
• AI-empowered (learner-as-leader): AI is used to empower learning while learners take agency

Table 1. Three paradigms of AIEd

Paradigm	Theoretical underpinning	Implementations	AI techniques
1. AI-Directed	Behaviorism	Intelligent Tutoring Systems (ITSs): software that tracks students' work, adjusts feedback and provides hints along the way)	AI based on statistical relational techniques
2. AI-supported	Cognitive, social constructivism	Dialogue-based Tutoring Systems (DTSs); Exploratory Learning Environments (ELEs)	Bayesian network, natural language processing, Markov decision trees
3. AI-empowered	Connectivism, Complex adaptive system	The human-computer cooperation; Personalized/adaptive learning	The brain-computer interface, machine learning, deep learning

P. Lameras and S. Arnab explored and analyzed what Artificial Intelligence means in Education (AIED) in their article "Power to the Teachers: An Exploratory Review on Artificial Intelligence in Education". The main take-away is that adaptivity and personalization are the innovation that AIED can offer to help students to learn and develop skills that are relevant to their own needs and experiences. However, it is important to help teachers to develop necessary digital competencies and skills for using AIED applications and tools in ethical and informed ways to enhance the student learning experience and attainment of learning outcomes. The findings of this review can contribute to developing a better understanding of how artificial intelligence may enhance teachers' roles as catalysts in designing and visualizing AI-enabled learning. As a result, more useful AI-systems specialized in pedagogy will be developed.

As Artificial Intelligence is becoming more apparent in education, it changes the way of teaching. Craig Coates, an entomologist at Texas A&M University, combated the cheating by reorienting his course towards writing and discussion, combined with adaptive courseware, also referred to as intelligent tutoring systems; and by using a tool that uses algorithms and analytics to grade submissions, the switch was a success. Advocates say this lets students study at their own pace and frees up the instructor’s time in class to shore up students’ knowledge.

The study conducted by S. Yang and H. Bai "The integration design of artificial intelligence and normal students’ Education" outlines four major problems of normal education (school for training teachers) development and explores the idea of using artificial intelligence technology to solve them. The main issues and the proposed solutions are:

Problem	Solution
Lack of teaching practice experience	Intelligent tutors/robots: it can make a special learning plan for students according to their interests, habits and learning needs. It can also be a simulation student of upcoming teachers to gain experience.
Outmoded curriculum	Expert system: can accurately classify normal students and highlight personalized teaching.
Lack of independent learning	Deep learning system: artificial intelligence can help to analyze the data of learners, organize the content according to the data, and carry out intelligent push. Teachers can create teaching resources to meet the needs of curriculum design and students.
Lack of innovation in teaching methods	Combination of all of the above.

Artificial intelligence and natural language processing systems are becoming increasingly prevalent in various forms of writing. These natural language processing (NLP) systems are trained on large databases of text to produce and refine statistical models that generate natural language responses. However, NLP systems do not "know" the meaning of the text they generate and can make factual and reasoning mistakes, which raises issues of accountability for authors using these systems. The use of NLP systems also raises questions of transparency in regards to authorship credit and contributions, and must be acknowledged in the text and references section of the manuscript.

The first chapter of this book aims to determine the knowledge and skills that remain for humans in an AI-driven world. It addresses AI's impact on education and highlights the need to develop awareness and understanding of AI. The goal is to create an educational environment where AI supports learners and teachers and prepares students for an AI-dominant future.

This article discusses recent developments in policies for medical journal publishing and editing in Korea. The article highlights that many editors in Korea are also publishers and therefore need to keep up with current trends and policies in the industry. The article focuses on six main policies that have emerged, including the use of artificial intelligence tools in publishing, preprint publications, open peer review, model text recycling policies, the updated 4th version of the Principles of Transparency and Best Practice in Scholarly Publishing, and the recommendation to include country names in human studies titles. The article also mentions that the use of AI in writing has increased, including in peer review and plagiarism detection. This shows the current and expected relevance AI has in modern scientific writing.

OpenAI launched a preview version of ChatGPT. It is part of the GPT (Generative Pre-trained Transformer) technology that can generate human-like text based on internet data. The release of ChatGPT sparked discussions on the impact of AI on education, with some saying it could render the student essay obsolete and others expressing concerns about cheating and false information. Despite its promise of transforming education, ChatGPT was criticized for its lack of understanding of meaning and content and its association with environmental racism and the interests of tech elites. Some educators called for better assignments that can't be written by an algorithm and for teaching students to use AI ethically. Access to ChatGPT was blocked by some US education departments due to concerns of dependence on technology, loss of privacy, and misuse of AI. This article opens up a range of important issues that are relevant across the educational field.

This paper provides a review of AI applications in higher education, with most of the research being quantitative research from computer science and STEM fields. So this review focuses more on AI utility over the entire academic institute rather than academic integrity and utility by students. Four areas of application were identified: profiling and prediction, assessment, adaptive systems and personalization, and intelligent tutoring systems. The conclusion highlights the need for further exploration of ethical and educational approaches in AI applications in higher education.

In the paper “OpenAI ChatGPT Generated Results: Similarity Index of Artificial Intelligence (AI) Based Model“, university director R. Ventayen researched how new AI powered technology enables cheating in the academic community. Since constructing an entire paper or essay using ChatGPT is considered a violation of academic integrity, the chatbot’s ability to pass similarity tests was researched on prompts with the headlines of the university’s publications. Their results show that ChatGPT provides acceptable similarly indices, and fails to provide the same citations or sources as the similar papers with the prompted headlines. Moreover, Quillbot paraphrases more content, which t, which provides plagiarism-free content.

This paper focuses foremost on the irreversibility of GPT-3 and its answers and ideas. It shows its shortcomings in dealing with semantic, mathematical and ethical questions and consequences for the writing industry. E.g, the big amount of written things that these tools can provide far exceeds the limited physical storage for them and marketing will abuse these tools heavily. We need to be critical with what these tools produce and how we use them. For our question regarding academic integrity, the problems with correct citations become apparent and academic texts produced by ChatGPT are often irreversibly produced from ideas of researchers.

This article shows how new developments in AI can solve 81% of university-level mathematics questions and even come up with new questions. This poses great advantages to the acquisition of new learning materials but the implications of this also addresses significant pedagogical challenges, with these models being available to students as well.

The aim of the article is to explore the academic and administrative applications of Artificial Intelligence. Artificial Intelligence Applications (AIA) are not only assisting education academically and administratively but also enhance their effectiveness. AIA provides help to teachers in various types of tasks in the shape of Learning Analytics (LA), Virtual Reality (VR), Grading/Assessments (G/A), and Admissions. It minimizes the administrative tasks of a teacher to invest more in teaching and guiding students. AIA adds a significant contribution to enhance student learning, minimize the workload of a teacher, grade/assess the students effectively and easily, and to help in a lot of other administrative tasks.

This report is an overview of research on AI applications in higher education between 2007 and 2018 through a systematic review. The combination of results gives us four areas of AIEd applications in academic support services, and institutional and administrative services: 1. profiling and prediction, 2. assessment and evaluation, 3. adaptive systems and personalisation, and 4. intelligent tutoring systems. The conclusions reflect on the almost lack of critical reflection of challenges and risks of AIEd, the weak connection to theoretical pedagogical perspectives, and the need for further exploration of ethical and educational approaches in the application of AIEd in higher education.

This article presents the focus of latest research in AIEd on reducing teachers’ workload, contextualized learning for students, revolutionizing assessments and developments in intelligent tutoring systems. It also discusses the ethical dimension of AIEd and the potential impact of the Covid-19 pandemic on the future of AIEd’s research and practice.

It investigates educational implications of emerging technologies on the way students learn and how institutions teach and evolve. Pinpointing some challenges for institutions of higher education and student learning in the adoption of these technologies for teaching, learning, student support, and administration and explore further directions for research.

This article anaylises a sample of 132 articles published betwee 2007-2017 about the scientific production on artificial intelligence in higher education. It concludes that, although artificial intelligence is a reality, the scientific production about its application in higher education has not been consolidated. Honestly not that interesting.

The paper is about a pedagogical experiment in which undergraduates were assigned to "cheat" by using text-generating AI software (GPT-2) to write their final class essay. The students were asked to reflect on the ethics of using AI in this way, such as what counts as plagiarism, and how working with AI could change their perspectives on writing, authenticity, and creativity. The results showed that composing with GPT-2 opened up the students' perspectives on the ethical use and evaluation of language models, and that their insights on these issues were connected to broader conversations in the humanities about writing and communication. The author of the paper shares the students' experiences and reflections on the experiment.

Survey

https://forms.office.com/Pages/DesignPageV2.aspx?subpage=design&FormId=R_J9zM5gD0qddXBM9g78ZOciNcKFkhVJsg7I7H6kiF5UNzNXQzhaOEpWSjZJSDg1WUZBN1lWTkkwUS4u

Appendix

Logbook

Week	Name	Total	Breakdown
1	Famke	8.5h	Group discussion (2h), Target group and requirements (1/2h), Studied papers (4h), Wrote summary for papers (2h)
	Gabriëlle	7h	Group discussion (1h), Studied papers (4h), Wrote summary for papers(1.5h), Target group and requirments(1/2h)
	My	8.5h	Group discussion (2h), Setup of wiki (0.5h), Studied papers (4h), Wrote summary for papers (2h)
	Naud	9h	Catch up (1h), Gathering information for appeal ERB (2h), Gathering relevant papers (1h), Studied papers (4h) Writing summary of papers (1h)
	Niels	8h	Group discussion (2h), Studied papers (3.5h), Wrote summary for papers (1.5h), Made Gantt chart (0.5h), made first agenda (0.5h)
	Quincy	7.5h	Group discussion (2h), Introduction & Problem statement and objectives (1/2h), studied papers (4h), wrote summary for papers (1h)
2	Famke	5.5h	Group discussion (1h), Target group and requirements (1/2h), Searched for literature (2h), Wrote summary for papers (2h)
	Gabriëlle	7h	Group discussion (1h), Searched for literature (1/2h), Studied papers (3h), Wrote summary for papers (1.5h), Group discussion (1/2h), Prepared for tutor meeting(1/2h)
	My	7h	Group discussion (1h), Searched for literature (1/2h), Studied papers (2h), Wrote summary for papers (1h), Read and organized already studied literature (2.5h)
	Naud	4.5h	Group discussion (1.5h), Filled in ERB review form (3h)
	Niels
	Quincy
3	Famke	3h 20m	Tutor meeting (20 min), Group discussion (1h), Worked on survey (2h)
	Gabriëlle		Tutor meeting (20 min), worked out target group (40 min)
	My
	Naud		Group discussion (1h), Filled in ERB review form (1h), Changed research question and stakehoders (0.5h)
	Niels
	Quincy
4	Famke		Tutor meeting (20 min), Group discussion (2.5h), Working on state of the art (4h)
	Gabriëlle
	My
	Naud		Tutor meeting (20 min), Group discussion (2.5h), ERB form correspondence (0.5h), updating wiki with survey questions(1h)
	Niels
	Quincy
5	Famke
	Gabriëlle
	My
	Naud
	Niels
	Quincy

Survey questions

Research into ChatGPT

In order to review the potential of using ChatGPT in academic education, we would like to know your experiences with the technology as well as your opinion on it. We will focus on the course Engineering Design, a mandatory course where you go through a complete design process, from the first idea to the realization of the product. We will look at how ChatGPT can be used for this course.

This survey is conducted for the course Project Robots Everywhere (0LAUK0), and its results will be used to shape further investigation in academic regulations regarding AI. The results of the questionnaire will be anonymous and it will take approximately 5-10 minutes to complete.

Do you give consent to participate in this study?

• Yes
• No

Knowledge of ChatGPT

Have you heard of ChatGPT before?

• Yes
• No

Do you know the capabilities of ChatGPT?

• Yes
• No

Could you name some of these capabilities?

• Open question

To give you an idea of what ChatGPT is, please read this:

ChatGPT is a large language model developed by OpenAI. It stands for “Generative Pre-training Transformer” and uses deep learning technology to generate text and respond to questions in a human way.

The model is trained on a large amount of text data, making it capable of understanding context and semantics. This makes ChatGPT a popular choice for chatbots, virtual assistants and other applications where human interaction is required.

Some examples of what ChatGPT can do:

- Write emails

- Write essays and academic reports

- Write poetry and song lyrics

- Compose music

- Answer (test) questions and solve problems

- Generate lines of code based on a prompt

- Answer customer queries

Usage of ChatGPT

Have you used ChatGPT before?

• Yes
• No

How often have you used ChatGPT?

• Only once or twice
• Monthly
• Weekly
• Daily

For what have you used ChatGPT?

• Open question

How would you rate your experience with ChatGPT in general?

• Rating from 0 to 5

How would you rate your experience with ChatGPT as an academic tool?

• Rating from 0 to 5
  

ChatGPT in academic education

Have you ever used ChatGPT for academic purposes? If yes, for what course(s)?

• Open question

Would you say usage of ChatGPT should be considered your own original work (so no plagiarism)?

• Yes
• No

When should the use of ChatGPT be considerd as plagiarism?

• Open question

Looking back, would you think ChatGPT could be a good addition to the course engineering design?

• Yes
• No

Please, if possible, elaborate on why? (Elaboration on previous question)

• Open question

For what do you think ChatGPT CAN be used in the course engineering design?

• Open question

For what do you think ChatGPT SHOULD be able to be used in the course engineering design?

• Open question
  

Learning objectives of Engineering Design

How likely do you think you would be able to reach the following learning objectives of Engineering Design with the use of ChatGPT? (Very likely - Very unlikely)

• Execute a generic design process
• Formulate a design goal
• Define the functional and technical specifications
• Generate an elaborate list of realization possibilities for the different functions of the design
• Select a number of design concepts from an extensive list of realization possibilities
• Make a final design choice between a number of concepts
• Develop a detailed design that meets the specifications
• Develop and execute a test plan for the prototype
• Evaluate a prototype based on test results and give an advice for redesign
• Reflect on the design and on the design process
• Write a design report describing the design process, the foundation of the choices made and the evaluation of the delivered design

Additional Learning Objectives

How likely do you think you would be able to reach the following learning objectives with the use of ChatGPT? (Very likely - Very unlikely)

• Gaining insight in designing a simple computer program
• Giving and understanding simple mathematical proofs
• Understand how basic electrical and electronic circuits work

Feelings about ChatGPT

What are your feelings regarding ChatGPT? (Strongly disagree - Strongly agree)

• I see ChatGPT as having/could have had a positive influence on my engineering design project
• I would have seen ChatGPT as a usefull tool for my engineering design project
• I see ChatGPT as a shortcut for tedious tasks I would have/had to perform in the course engineering design
• I will use/would have used ChatGPT in the course engineering design regularly
• I feel like I would have learned something when using ChatGPT for the engineering design project
  

Interview script Teachers

Interviewee: [name] Date: [date]

Interviewer: [name] Location: [location]

Introduction

Hi, my name is ___ and I’m conducting this interview as part of our project for the course Project Robots Everywhere.

The goal is to review the potential of using ChatGPT, an AI chatbot, in academic education, so we would like to know your experiences with this technology as well as your opinion on it. We will mainly focus on the course Engineering Design and will look at how ChatGPT can be used for this course.

1. What is your name?
2. What is your function / role within Engineering Design? (lecturer, tutor, etc.)

Knowledge and experience

Before we go into the specific course, we’d like to know what your knowledge and experience is regarding ChatGPT

3. Have you heard of ChatGPT before?

4. Do you know its capabilities?

a. If yes, give examples

[Explain capabilities, show examples]

Some examples of what ChatGPT can do:

• Write emails
• Write essays and academic reports
• Write poetry and song lyrics
• Compose music
• Answer (test) questions and solve problems
• Generate lines of code based on a prompt
• Answer customer queries

5. Have you ever used ChatGPT before?

a. If yes, what for?

b. Have you ever used ChatGPT as an academic tool?

c. How would you rate ChatGPT as an academic tool?

d. How would you describe your overall experience with ChatGPT?

ChatGPT in academic education

Now we will go more in depth about the potential usage of ChatGPT in academic education, and more specifically, Engineering Design.

6. What is your opinion on using ChatGPT as an academic tool?

7. What is your expectation for the future of academic research with these AI technologies on the rise? Do you believe the course should change to reflect this?

8. How do you think chatbots like ChatGPT would influence students’ ability to write? Does this also affect their ability to write code?

9. Does using ChatGPT count as plagiarism in your opinion?

10. Would you be open to using / allowing your students to use ChatGPT in your course?

a. If yes, how would you implement it? Would you actively encourage it

b. If no, why?

11. Would the learning objective of Engineering Design still be obtainable when using ChatGPT?

a. Which ones will become “useless” since automation is becoming more generalized? (for example, writing simple code)

b. What would you change about the learning objectives? (make them more complex?)

c. How would you adjust the assignments to fit the new criteria?

12. What would you think of giving students an assignment in which they are required to use ChatGPT?

13. In your opinion, does ChatGPT make the education of students better, or does it hinder them in their development?

Survey input

t.b.d.

Theory

ChatGPT is a large language model developed by OpenAI. It stands for “Generative Pre-training Transformer” and uses deep learning technology to generate text and respond to questions in a human way.

Some examples of what ChatGPT can do:

• Write emails
• Write essays and academic reports
• Write poetry and song lyrics
• Compose music
• Answer (test) questions and solve problems
• Generate lines of code based on a prompt
• Answer customer queries

Possible questions

• How do you think chatbots like ChatGPT would influence students’ ability to write?
• What would you think of giving students an assignment in which they are required to use ChatGPT?
• Does using ChatGPT count as plagiarism in your opinion?
• In your opinion, does ChatGPT make the education of students better, or does it hinder them in their development?

Interview Script Student "Experts"

Interviewee:[name] Date:[date]

Interviewer:[name] Location:[location]

Introduction

Hi, my name is ___ and I’m conducting this interview as part of our project for the course Project Robots Everywhere.

The goal is to review the potential of using ChatGPT, an AI chatbot, in academic education. You have been selected because you are well known about ChatGPT and its uses, so we would like to know your experiences with this technology as well as your opinion on it.

1. How long have you been studying at the TU/e?
2. How long have you been using ChatGPT

Knowledge and experience

1. For what purposes do you use ChatGPT?
2. Why do you use ChatGPT?
3. For which of these purposes do you find ChatGPT works the best?
4. How do you feel about your experiences with ChatGPT?

ChatGPT in academic education

Now we will go more in depth about the potential usage of ChatGPT in academic education, and more specifically, Engineering Design.

1. What is your opinion on using ChatGPT as an academic tool?
2. Does using ChatGPT count as plagiarism in your opinion?
3. Did you complete the course Engineering Design?
4. Would you, and when yes, how would you use ChatGPT to obtain the following learning objectives of Engineering Design?
   1. Execute a generic design process
   2. Formulate a design goal
   3. Define the functional and technical specifications
   4. Generate an elaborate list of realization possibilities for the different functions of the design
   5. Select a number of design concepts from an extensive list of realization possibilities
   6. Make a final design choice between a number of concepts
   7. Develop a detailed design that meets the specifications
   8. Develop and execute a test plan for the prototype
   9. Evaluate a prototype based on test results and give an advice for redesign
   10. Reflect on the design and on the design process
   11. Write a design report describing the design process, the foundation of the choices made and the evaluation of the delivered design

1. Do you think you would actually learn these objectives while using ChatGPT?
2. In your opinion, does ChatGPT make the education of students better, or does it hinder them in their development?