Does ChatGPT-Permitted Assessments Help Students Generate

Better Answers and Learn More?

Michelle LF Cheong

and Jean Y-C Chen

School of Computing & Information Systems, Singapore Management University,

80 Stamford Road, Singapore 178902, Singapore

Keywords: ChatGPT 3.5, Permitted Use, Assessment, Higher Education, Empirical Study.

Abstract: We discuss our methodology and implementation of ChatGPT-permitted assessments for a university-level

spreadsheets modelling module. Through our quantitative data analysis, our students rated ChatGPT’s

answers to be incorrect on average and thus will not help them generate better answers directly, representing

low “Perceived usefulness” (PU), while they rated ChatGPT 3.5 with relatively high “Perceived ease of use”

(PE). They gave a good “Behavioural intention” (BI) rating indicating that they were motivated to use it in

future as they could still learn more about this module by using ChatGPT 3.5. We found that both PU and PE

affected BI positively, with PU being the stronger predictor, suggesting that developers should focus on

improving ChatGPT’s accuracy to improve PU, which will in turn have a higher positive impact on BI.

Through our qualitative analysis, our students indicated that they could learn positively from ChatGPT 3.5 in

terms of getting an initial idea on how to approach the problem, providing a first cut solution, learning the

execution steps for complex Excel functions, providing an active learning opportunity through identifying

and correcting the mistakes, and gaining the awareness of not committing such mistakes in the future.

1 INTRODUCTION

Since the launch of ChatGPT 3.5 in November 2022,

many stakeholders including students, instructors,

and institutions were amazed by its mostly accurate

human-like responses in a myriad of domain areas

including medical (Kung et al., 2023), journalism

(Pavlik, 2023) and programming (Anagnostopoulos,

2023). However, at the same time, many have

expressed concerns of academic integrity when

students submit ChatGPT responses as their own, and

excessive dependence may erode students’ writing

and critical thinking skills (Lim et al., 2023;

O’Connor and ChatGPT, 2023; Stokel-Walker,

2022). In response, institutions around the world

rushed to establish guidelines and policies on how to

handle this sudden “invasion” of GAI tools to soften

its negative impact on compromised education

quality and how to exploit it to achieve positive

impact on education (Haleem et al., 2022; Moorhouse

et al., 2023).

https://orcid.org/0000-0002-3192-3452

https://orcid.org/0000-0003-2338-4265

Chan (2023) conducted a survey with Hongkong

universities and proposed an AI Ecological Education

Policy Framework which is organized into three

dimensions: Pedagogical, Governance, and

Operational. In terms of pedagogical dimension, they

recommended “teachers to design assessments that

allow AI technologies to enhance learning outcomes,

rather than solely producing outputs” and to “focus

on students’ understanding, critical thinking, and

analysis to prevent AI-generated content from

compromising the assessment process”. Moorhouse

et al. (2023) examined the guidelines of 23 of the top

50 universities and they covered three common areas:

academic integrity, advice on assessment design, and

communication with students. For advice on

assessment design, they include testing the

assessments using GAI tools to understand the

abilities and limitations, redesigning assessment

tasks, focus on process and staged assessment design,

incorporating GAI tools in the assessment process,

and use them during in-class assessments.

Cheong, M. L. and Chen, J. Y.-C.

Does ChatGPT-Permitted Assessments Help Students Generate Better Answers and Learn More?.

DOI: 10.5220/0013095500003932

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 17th International Conference on Computer Supported Education (CSEDU 2025) - Volume 2, pages 189-197

ISBN: 978-989-758-746-7; ISSN: 2184-5026

189

The idea to incorporate GAI tools in the

assessment process where GAI tool usage is

permitted to aid with assessment completion is a

fairly new approach. To do so will require careful

design of the assessment, including the

implementation process detailing what the students

need to do and allowed to do when using GAI tools,

how to separate GAI versus student generated

answers, and therefore what will be graded and what

will not. In this paper, we proposed a methodology to

incorporate ChatGPT permitted use in assessment

completion for a university-level spreadsheets

modelling module. In this methodology, our students

used ChatGPT 3.5 as an assistant to complete their

assignments and created two separate spreadsheet

models for each question, one based purely on

answers suggested by ChatGPT and a second

improved version based on students’ own suggestions

for improvements, and we determined if ChatGPT

helped them generate better answers and whether they

learn more from using it, through analysing the

quantitative and qualitative survey data collected.

2 LITERATURE REVIEW

To address the concerns of academic integrity and

dishonesty, and decline in students’ writing and critical

thinking skills, many articles (Cheong, 2023; Lim et a.,

2023; O’Connor and ChatGPT, 2023; Zhai, 2022)

suggested instructors to redesign assessments to focus

on higher order thinking skills which are more “GAI-

resistant”. Instructors are also encouraged to include

ChatGPT permitted use as part of the assessment

process since GAI will likely become a “reality of

today’s educational and job landscape” (Halaweh,

2023; Markauskaite et al., 2022; Moorhouse et al,

2023). Specifically, Halaweh (2023) presented

arguments in favour of it and proposed five strategies

and techniques to ensure responsible and successful

implementation of ChatGPT in teaching and research.

These include setting clear policy for ChatGPT usage;

requiring students to document the steps in the usage

including contradictory findings, judgments and

improvements made, and complete audit trails of

questions asked, and corresponding answers received;

use AI detector tools to inform contents similarity and

plagiarism; and finally, instructor and student to swap

roles to evaluate the authenticity of learning and critical

thinking.

However, there are fewer articles on actual

implementation of redesigned assessments and GAI-

incorporated assessments with students. In French et

al., (2023), the authors also commented that there are

many commentaries on advantages and limitations of

GAI in education and proposed recommendations,

but there are “few examples of actual practice with

students”. They conducted a study to integrate

ChatGPT and Dall-E into a research and development

assignment with their games programming students.

Their students evaluated the tools in the context of

game development, demonstrated working

prototypes integrating ChatGPT and Dall-E, and

reported on their findings. They discussed five

student outputs in detail, highlighting the students’

learnings, frustrations they had, and what ChatGPT

and Dall-E can and cannot do well in games

development. However, no statistical data analysis of

student feedback was performed.

Another actual implementation is by Polasik

(2023) who asked her students to use ChatGPT in

their learning process in an introductory materials

science course. These include getting the students to

ask ChatGPT to generate a list of conceptual

questions to ask among themselves to discover gaps

in understanding, ask ChatGPT to explain the main

themes in their own reports, and ask ChatGPT to

write MATLAB scripts to accomplish small tasks.

Again, no statistical data analysis of student feedback

was performed .

Ngo (2023) conducted a survey with 200 students,

and a semi-structured interview with 30 students to

investigate how Vietnamese university students

perceived the use of ChatGPT in education and

provided recommendations to overcome the usage

challenges. Descriptive statistics and one-sample t-

tests were conducted, and it was found that students’

positive perception of use and educational advantage

were both higher than average. The students also had

above average awareness of the challenges associated

with the use of ChatGPT. It is worthwhile to note that

the survey results were related to general usage of

ChatGPT in education, and not specifically in

assessment completion.

Our work contributes towards the limited

empirical research in actual implementation of

ChatGPT-permitted assessments in higher education.

Our work is different from earlier works in three

aspects. Firstly, our work was based on assessments

in a spreadsheets modelling module where students

will build mathematical models and perform complex

calculations to answer business questions. To our best

knowledge, we could only find one recent work by

Cheong (2023) that discussed the performance of

ChatGPT 3.5 on spreadsheets modelling assessments,

to determine how well ChatGPT 3.5 can solve

questions of different cognitive levels based on the

revised Bloom’s Taxonomy.

CSEDU 2025 - 17th International Conference on Computer Supported Education

190

Secondly, as suggested by Moorhouse et al.,

(2023), “allowing or even requiring students to use

GAI at various stages of the assessment process

would, in fact, enhance the authenticity of

assessments”. We required our students to use

ChatGPT 3.5 to generate answers to the assignment

questions and prepare a complete audit trail of the

questions asked and answers received as

recommended by Halaweh (2023). In addition,

students must identify and document down mistakes

committed by ChatGPT 3.5, following the

recommendation by Chan (2023) to use “assessments

and activities where students can by themselves

discover the limits of such techniques”, and by

Halaweh (2023) to get students to document down the

contradictions. After the mistakes were identified, the

students must generate an improved version of the

answers based on their own suggestions for final

submission and grading. Our proposed methodology

of incorporating ChatGPT as an assistant in

assessment completion with full audit trail and

identification of mistakes and learnings at each step,

and then create an improved solution model, presents

an active learning model where students build the

ability to critically judge the quality of responses

generated by ChatGPT to develop evaluative

judgement skill (Bearman et al., 2024).

Thirdly, our actual implementation allowed us to

collect both quantitative and qualitative survey data

from our students on the entire assessment

completion process, to assess the efficacy of using

ChatGPT 3.5 in assisting them to complete their

assignments and determine what they learn from

using it. With our data analysis, we aim to answer our

research questions:

 RQ1: Does the usage of ChatGPT 3.5 help

students to generate better answers for the

assignment questions?

 RQ2: Does the usage of ChatGPT 3.5 help

students to learn more in this module?

 RQ3: What are the learnings in using ChatGPT

3.5 to complete the assignment questions?

3 MODULE AND ASSESSMENT

3.1 Spreadsheets Modelling Module

The spreadsheets modelling module teaches students

how to translate business problems into mathematical

representations, and to build the spreadsheet models

from scratch to perform mathematical calculations

and data analysis to obtain insights for decision

making. The module covers six main topics, namely:

(i) basic modeling techniques, (ii) spreadsheets

engineering skills, (iii) financial calculations, (iv)

data lookups and optimization, (v) Monte Carlo

simulation, and (vi) time-based discrete event

simulation.

3.2 Assessment Design and

Implementation

We designed two take-home assignment questions

(denoted as AQ1 and AQ2) for the fall term of

academic year 2023/24. As the assignment questions

were original and did not exist before September

2021, they will not form part of the ChatGPT’s

training corpus, hopefully making them more “GAI-

resistant”. Each question described a business

scenario and contained multiple consecutively linked

parts for end-to-end analysis.

For each question, students will complete it

following the steps below as depicted in Figure 1.

1) Feed the question part by part as prompts to

ChatGPT 3.5 in a continuous conversation, to

obtain suggested answers to each part. Create the

spreadsheet model (Model A) according to the

suggested answers generated by

ChatGPT 3.5.

Rate if the student can learn more

about this module by using ChatGPT

3.5 in the learnin

rocess.

Rate if the student is engaged and

motivated to use ChatGPT 3.5 in

future learnin

in this module.

We computed Cronbach’s alpha (Taber, 2018) to

determine our students’ overall consistency in terms

of their responses across all three factors for each

questionnaire, and obtained the value of 0.8724 for

Questionnaire 1, and 0.8389 for Questionnaire 2,

indicating that both are in the good range. As the

number of items in the questionnaire was different for

each factor, we also computed Spearman-Brown

Coefficient (Eisinga et al., 2013) for each factor for

both questionnaires in Table 2. Most of the

coefficients can be interpreted as excellent, except for

the factor PE which were Fair for Questionnaire 1,

and Good for Questionnaire 2. Overall, these values

indicated that our students’ responses were

consistent, and we could proceed to analyse their

responses to gain insights.

Table 2: Spearman-Brown Coefficients.

Facto

earman-Brown Coeff Inter

retation

Q1: PU

0.940 Excellen

Q1: PE

0.625 Fai

Q1: BI

0.917 Excellen

Q2: PU

0.912 Excellen

Q2: PE

0.737 Goo

Q2: BI

0.917 Excellen

We performed one-tail test for mean scores above

5.0 (out of 10) for questions 1 and 2, and mean scores

above 2.5 (out of 5) for questions 3 to 6, to obtain the

p-value to determine significance.

For Questionnaire 1, the mean scores for

Questions 1a to 1d, and 2a to 2d given in Table 3 were

mostly below the average of 5.0, except for Question

1a with mean score of 5.238 but was found to be

Does ChatGPT-Permitted Assessments Help Students Generate Better Answers and Learn More?

193

insignificant with p-value of 0.15. This indicates that

on average, students did not rate ChatGPT’s answers

to be correct and thus did not help them generate

better answers directly, representing low “Perceived

usefulness” (PU). For questions 3 and 4, the mean

scores were 3.905 and 2.968 respectively, and both

had very low p-values. This indicates that on average,

students rated ChatGPT 3.5 with relatively high

“Perceived ease of use” (PE). Finally, for questions 5

and 6, the mean scores were 2.984 and 3.056

respectively, and both with very low p-values. This

indicates that on average, students felt that they can

learn more from ChatGPT 3.5 and were engaged and

motivated to use it for future learning in this module,

representing good “Behavioural intention” (BI).

Table 3: Questionnaire 1 ratings.

Factor Mean

score

SD p-value

1a PU

5.238 2.559 0.15

1b PU

3.579 2.405 -

1c PU

3.413 2.338 -

1d PU

3.127 2.257 -

2a PU

4.865 2.521 -

2b PU

3.690 2.425 -

2c PU

3.500 2.510 -

2d PU

3.317 2.493 -

3 PE

3.905 0.971 9.89 x 10-33

4 PE

2.968 1.069 1.46 x 10-6

5 BI

2.984 1.016 2.22 x 10-7

6 BI

3.056

1.086 3.73 x 10-8

For Questionnaire 2, the results given in Table 4

were similar to that of Questionnaire 1, with a slight

difference for question 1a where the mean score of

5.794 was found to be significant. The overall results

for Questionnaire 2 were low PU, relatively high PE,

and good BI, identical to that of Questionnaire 1.

Table 4: Questionnaire 2 ratings.

Factor Mean

score

SD p-value

1a PU

5.794 3.252 0.0036

1b PU

3.548 2.080 -

1c PU

2.587 1.912 -

2a PU

4.841 3.030 -

2b PU

3.579 2.220 -

2c PU

2.563 2.026 -

3 PE

3.563 1.137 4.58 x 10-19

4 PE

2.921 0.997 3.14 x 10-6

5 BI

2.897 1.060 2.67 x 10-5

6 BI

2.944 1.064 3.81 x 10-6

This is an interesting finding, as it implies that

while our students did not find ChatGPT’s answers

accurate and useful most of the time, they were still

able to learn something from it and were motivated to

use it in the future. This could be due to a few possible

reasons. One, the relatively high PE drives BI

positively; two, the process of identifying and

correcting the mistakes enhance learning and students

found such active learning useful for them; and three,

the sheer intrinsic motivation to use new and exciting

tools.

We performed a multiple linear regression to

understand the impact of PU and PE on BI, by

combining the survey responses of both

questionnaires. The relationship obtained was BI =

0.5261*PU + 0.2463*PE + 0.2294, and the p-values

were all very low and thus the coefficients and

constant were all significant, as given in Table 5. Our

results were similar to that of Duong et al. (2023)

where both PE and PU had positive impact on BI. In

addition, PU was a stronger predictor than PE,

suggesting that if the developers can focus on

improving ChatGPT’s accuracy to improve PU, it

will in turn have a higher positive impact on BI,

which was also recommended by Lai et al. (2023).

Table 5: Multiple linear regression of PU and PE on BI.

Coeff Std Erro

>|t|

Cons

0.2294 0.070 3.275 0.001

PU 0.5261 0.102 5.135 0.000

PE 0.2463 0.101 2.444 0.016

4.2 Qualitative Data Analysis

Based on our students’ qualitative descriptions of the

wrong answers and the areas which they have learnt

positively from the responses provided by ChatGPT

3.5, we performed text mining and created word

cloud and topic modelling using Latent Dirichlet

Allocation (LDA) for each part of AQ1 and AQ2.

For example, for AQ1 part a, the question required

the students to compute the present value worth of an

investment which pays dividends at the end of 5th,

6th, 7th, 8th and 9th year, using a return rate of 5%.

The word cloud (Figure 2) for the feedback on the

wrong answers provided by ChatGPT 3.5 highlighted

that it understood the year wrongly as it

misinterpreted the information that no dividends will

be returned in the first four years and assumed year 5

to be year 1 thus leading to wrong values entered, plus

the suggested Excel model had confusing cell

locations.

The corresponding topic model is given as

0.053*year + 0.028*cell + 0.025*value, representing

CSEDU 2025 - 17th International Conference on Computer Supported Education

194

the most committed mistake by ChatGPT 3.5

involving year, cell and value. Specific student

comments include:

“Did not understand that the dividend is in year

5-9, so the calculations are all incorrect.”

“Wrong location of year cells which led to wrong

input of dividends.”

Figure 2: Word cloud for wrong answers to AQ1 part a.

Using another example, in AQ2 part b, the

question required the students to create an Excel

model to simulate a boat race to determine the winner

between two fictitious teams, Oxfort and Cambrick,

using random generators and probability distribution

functions. The word cloud (Figure 3) for the feedback

on the wrong answers provided by ChatGPT 3.5

highlighted that it was not able to apply the correct

formula to perform Monte Carlo simulation,

especially for team Oxfort, and thus unable to

determine the winner.

Figure 3: Word cloud for wrong answers to AQ2 part b.

The corresponding topic model is given as

0.042*formula + 0.018*winner + 0.018*oxfort,

representing the most committed mistake by

ChatGPT 3.5. Specific student comments include:

“The formula to calculate Oxfort position for

each minute did not work.”

“Winner of the race is not calculated properly,

and the winner column is entirely wrong.”

From these two examples, we can see that

ChatGPT 3.5 can misinterpret the information

provided in the question and was unable to apply the

correct formula to perform some calculations. Such

an outcome was in fact an expected one, given that

ChatGPT 3.5 is primarily a language model and does

not fare as well in mathematical calculations, even

more so in spreadsheets modelling. This phenomenon

was similarly highlighted in Cheong (2023)

particularly for questions of higher cognitive levels.

Albeit these mistakes committed by ChatGPT 3.5,

the students provided comments on what they have

learnt positively from it. For example, for AQ1 part

c, the question required students to use iterative

method to determine the new dividend value for a

desired return rate. The word cloud (Figure 4) shows

that students learnt how to use Goal Seek to perform

the iterative calculations following the steps

suggested by ChatGPT 3.5 to find the answer.

Figure 4: Word cloud for positive learning to AQ1 part c.

The corresponding topic model is given as

0.061*seek + 0.059*goal + 0.035*use, representing

the positive learning from ChatGPT 3.5. Specific

student comments include:

“Using of the goal seek function from the

instructions provided by GPT.”

“ChatGPT 3.5 was correct in pointing out that the

Goal Seek function can be used to find the dividend

amount based on the return rate of 8%.”

Quoting another example, in AQ2 part c, the

question required the student to use Data Table to

manage the simulation of 30 such races and compute

the winning probability of team Oxfort from the 30

simulated races. The word cloud (Figure 5) shows

that students learnt how to use Data Table to repeat

the simulation 30 times following the steps suggested

by ChatGPT 3.5.

The corresponding topic model is given as

0.065*table + 0.064*data + 0.029*use, representing

the positive learning from ChatGPT 3.5. Specific

student comments include:

“How to navigate Data Table. The instructions

were very clear.”

“It realizes that it needs to set up the format of a

data table and it gives the steps very accurately just

Does ChatGPT-Permitted Assessments Help Students Generate Better Answers and Learn More?

195

that it does not specify which cell or which formula,

so it gives a high level idea but it may not be totally

accurate.”

Figure 5: Word cloud for positive learning to AQ2 part c.

Such an outcome was similarly highlighted in

Cheong (2023) that ChatGPT 3.5 was very capable in

explaining how to use Excel functions correctly by

providing clear step-by-step instructions, while it may

not be able apply them correctly all the time.

However, not all students provided comments on

positive learning from ChatGPT 3.5. Some students

explicitly stated that they learnt nothing from

ChatGPT 3.5 (6.0% for AQ1 and 15.4% for AQ2). If

we include students who indicated “NA” for their

responses, we obtained 21.2% for AQ1 and 32.6% for

AQ2. These percentages were not high, in

comparison with students who explicitly described

their positive learnings.

4.3 Discussion of Overall Results

Overall, our students’ responses highlighted that

while ChatGPT 3.5 was very easy to use (high PE),

the quality of its generated responses to the

spreadsheets modelling questions were low (low PU).

However, the students were still motivated to use

ChatGPT 3.5 in future for learning this module (good

BI), as there were some “nuggets of wisdom” which

they could glean from its responses. Some of the

positive students’ comments include:

“Provided some ideas of how to approach the

problem.”

“Give me the framework on how to solve the

problem, although there are different corrections that

need to be modified to get the correct answer.”

“ChatGPT’s model gave me a starting point to

structure the model, as well as alerted me to

mistakes/inputs that I should not make for the

calculations to be valid.”

With our data analysis results, we attempt to

answer our research questions. For RQ1, since the

quality of its generated responses to the spreadsheets

modelling questions were low (low PU), ChatGPT

3.5 will not be able to help students generate better

answers directly. However, it does provide an initial

idea of how to approach the question and provide a

first cut solution for the students to improve upon,

which will assist the students indirectly.

For RQ2, our students felt that they can learn more

from ChatGPT 3.5 and were engaged and motivated

to use it for future learning (good BI). For RQ3, the

learnings from ChatGPT 3.5 include the execution

steps for complex Excel functions, active learning

through identifying and correcting the mistakes

committed by ChatGPT 3.5, and gaining the

awareness of not committing such mistakes in the

future. Such learnings form part of developing

evaluative judgement skill in the students which

aligns with the learning outcomes of the module.

5 CONCLUSIONS AND FUTURE

WORK

It is expected that the quality of the GAI-generated

responses will improve over time, and GAI tools will

become embedded in common software tools such as

MS office tools. Allowing and incorporating GAI

tools usage in educational assessments will become

inevitable. To be better prepared, instructors must

engage in purposeful assessment designs that allow

students to exploit the GAI tools to generate better

solutions to real world problems, and at the same be

cognizant about the strengths and weaknesses of such

tools, and to use them responsibly.

Our methodology to incorporate ChatGPT in

assessment completion with complete audit trails,

identifying and documenting the mistakes made and

positive learnings, and suggesting improvements,

allows students to learn actively from the entire

process developing better evaluative judgment skill

and achieving better learning outcomes. Other

instructors can adopt our methodology in their own

course assessments to obtain their specific findings.

The proposed future work that should follow

immediately would be to identify and quantify

students’ skills, abilities and knowledge gain in using

GAI tools in learning, by comparing with a control

group that does not use GAI tool. In addition, we can

also collect and analyse data on students’ and

instructors’ perception of whether using such tools in

learning will indeed result in the decline of students’

critical thinking skills or increase students’ reliance on

these tools. The results will serve to validate some of

the common concerns that were raised in many articles.

CSEDU 2025 - 17th International Conference on Computer Supported Education

196

For course instructors, one future work would be to

establish a methodology on testing their assessments

using GAI tools to understand the abilities and

limitations to inform assessment redesign to construct

more effective assessments that can develop the

desired skills for the next-generation workforce.

REFERENCES

Anagnostopoulos, C-N. (2023). ChatGPT impacts in

programming education: A recent literature overview

that debates ChatGPT responses. F1000Research 2023.

https://doi.org/10.12688/f1000research.141958.1

Bearman, M., Tai, J., Dawson, P., Boud, D., Ajjawi, R.

(2024). Developing evaluative judgement for a time of

generative artificial intelligence. Assessment &

Evaluation in Higher Education, 49:6, 893-905.

https://doi.org/10.1080/02602938.2024.2335321

Chan, C.Y.K. (2023). A comprehensive AI policy

education framework for university teaching and

learning. International Journal of Educational

Technology in Higher Education, 20:38.

https://doi.org/10.1186/s41239-023-00408-3

Cheong, M. (2023). ChatGPT’s Performance in

Spreadsheets Modelling Assessments based on Revised

Bloom’s Taxonomy. In Proceedings of the 31st

International Conference on Computers in Education

(ICCE 2023). Asia-Pacific Society for Computers in

Education.

Duong, C.D., Bui, D.T., Pham, H.T., Vu, A.T., Nguyen,

V.H. (2023). How effort expectancy and performance

expectancy interact to trigger higher education

students’ uses of ChatGPT for learning. Interactive

Technology and Smart Education, 1741-5659.

https://doi.org/10.1108/ITSE-05-2023-0096

Eisinga, R., Grotenhuis, M., Pelzer, B. (2013). The

reliability of a two-item scale: Pearson, Cronbach, or

Spearman-Brown? International Journal of Public

Health, 58(4), 637-642. https://doi.org/10.1007/

s00038-012-0416-3

French, F., Levi, D, Maczo, C., Simonaityle, A., Triantafyl-

lidis, S, Varda, G. (2023). Creative Use of OpenAI in

Education: Case Studies from Game Development.

Multi-modal Technologies and Interaction. 7, 81.

Halaweh, M. (2023). ChatGPT in education: Strategies for

responsible implementation. Contemporary

Educational Technology, 15(2), ep421. https://doi.org/

10.30935/cedtech/13036

Haleem, A., Javaid, M., Singh, R.P. (2022). An era of

ChatGPT as a significant futuristic support tool: A

study on features, abilities, and challenges.

BenchCouncil Transactions on Benchmarks, Standards

and Evaluations 2:100089. https://doi.org/10.1016/j.

tbench.2023.100089

Kung, T.H., Cheatham, M., Medenilla, A., Sillos, C., De

Leon, L., Elepano, C., Madriaga, M., Aggabao, R.,

Diaz-Candido, G., Maningo, J., Tseng, V. (2023).

Perfor-mance of ChatGPT on USMLE: Potential for

AI-assisted medical education using large language

models. PLOS Digit Health, 2(2): e0000198.

https://doi.org/10.1371/journal.pdig.0000198

Lai, C.Y., Cheung, K.Y., Chan, C.S. (2023). Exploring the

role of intrinsic motivation in ChatGPT adoption to

support active learning: An extension of the technology

acceptance model. Computers and Education: Artificial

Intelligence, 5:100178. https://doi.org/10.1016/j.caeai.

2023.100178

Lim, W.M., Gunasekara, A., Pallant, J.L., Pallant, J.I.,

Pechenkina, E. (2023). Generative AI and the future of

education: Ragnarok or reformation? A paradoxical

perspective from management educators. The

International Journal of Management Education, 21,

100790. https://doi.org/10.1016/j.ijme.2023.100790

Markauskaite L., Marrone R., Poquet O., Knight S.,

Martinez-Maldonado R., Howard S., Tondeur J., De

Laat M., Buckingham S.S., Gasevic D., Siemens G.

(2022). Rethinking the entwinement between artificial

intelligence and human learning: what capabilities do

learners need for a world with AI? Computers &

Education: Artificial Intelligence. 3:100056.

https://doi.org/10.1016/j.caeai.2022.100056

Moorhouse, B.L., Yeo, M.A., Wan, Y. (2023). Generative

AI tools and assessment: Guidelines of the top-ranking

universities. Computers and Education Open 5:100151.

https://doi.org/10.1016/j.caeo.2023.100151

Ngo, T.T.A. (2023). The perception by university students

of the use of ChatGPT in education. International

Journal of Emerging Technologies in Learning (iJET),

18(17), 4–19. https://doi.org/10.3991/ijet.v18i17.39019

O’Connor, S., ChatGPT. (2023). Open artificial

intelligence platforms in nursing education: Tools for

academic progress or abuse? Nurse Education in

Practice, 66.

Pavlik, J.V. (2023). Collaborating with ChatGPT:

Considering the implications of Generative Artificial

Intelligence for journalism and media education.

Journalism & Mass Communication Educator, 78(1),

84-93. https://doi.org/10.1177/10776958221149577

Polasik, A.K. (2023). Hey ChatGPT: Can you help me

learn? The Journal of The Minerals, Metals &

Materials Society (TMS), 75(7), 2089-2090.

https://doi.org/10.1007/s11837-023-05924-1

Stokel-Walker, C. (2022). AI bot ChatGPT writes smart

essays — should professors worry? Nature.

https://doi.org/10.1038/d41586-022-04397-7.

Strzelecki, A. (2023). To use or not to use ChatGPT in

higher education? A study of students’ acceptance and

use of technology. Interactive Learning Environments.

https://doi.org/10.1080/10494820.2023.2209881

Taber, K.S. (2018). The use of Cronbach’s Alpha when

developing and reporting research instruments in

Science education. Research in Science Education, Vol

48, 1273-1296. https://doi.org/10.1007/s11165-016-

9602-2

Zhai, X. (2022). ChatGPT user experience: Implications

for education. https://papers.ssrn.com/sol3/papers.

cfm?abstract_id=4312418.

Does ChatGPT-Permitted Assessments Help Students Generate Better Answers and Learn More?

197