Practices, Challenges, and Training Needs of Faculty in Terms of

Generative AI

Britt Petjärv

, Vitali Retsnoi

, Anne Uukkivi

, Monica Vilms

, Elena Safiulina

and Oksana Labanova

Centre for Humanities and Economics, TTK University of Applied Sciences, Pärnu mnt 62, Tallinn, Estonia

Centre for Sciences, TTK University of Applied Sciences, Pärnu mnt 62, Tallinn, Estonia

Institute of Engineering and Circular Economy, TTK University of Applied Sciences, Pärnu mnt 62, Tallinn, Estonia

Keywords: Generative AI (GenAI), Faculty Perceptions, Higher Education, Training Needs, Academic Integrity.

Abstract: This paper investigates the role of Generative AI (GenAI) tools in higher education at TTK University of

Applied Sciences (TTK UAS), Estonia, the largest applied sciences university in Estonia providing higher

education in engineering. Through a survey of 81 faculty members, it examines the use of GenAI in teaching,

research, and administrative tasks, highlighting patterns of usage, perceived benefits, challenges, and training

needs. The findings reveal that while GenAI is seen as a valuable asset in personalized learning and efficient

task management, concerns about reliability, ethical implications, and workload dynamics persist. The study

emphasizes the importance of targeted training to address these challenges and support the effective

integration of GenAI tools in higher education.

1 INTRODUCTION

Generative AI (GenAI) is a type of artificial

intelligence (AI) technology capable of generating

new and unique outputs, such as images, text, audio,

videos, 3D models (Holmes & Miao, 2023).

Due to

its ability to produce sophisticated and realistic

content that reflects human creativity, GenAI has

become a valuable tool in various industries,

including education, entertainment, and product

design (Castelli & Manzoni, 2022).

Since the launch of ChatGPT (Chat Generative

Pre-trained Transformer - the fastest-growing app in

history to date, based on large language models) at the

end of 2022, extensive discussions and widespread

research have emerged, raising both concerns and

innovative ideas for enhancing higher education

(

Holmes & Miao, 2023). Based on studies and general

trends published over the past two years, it can be

stated that AI technologies have the potential to

significantly transform teaching and learning in

higher education (Holmes & Miao, 2023

; Ward et al,

2024).

However, several challenges remain, including

addressing ethical and quality concerns within higher

education, navigating implementation issues, and

utilizing appropriate pedagogical frameworks. To

best support faculty members at TTK UAS in

navigating these challenges, it was necessary to first

map out the existing knowledge, attitudes, and

practices related to AI. This foundation would allow

for targeted training to address the identified gaps.

More specifically, the aim of the study is to determine

the current use of GenAI tools in both the planning

and delivery of teaching, as well as in other activities

associated with faculty work (such as projects,

research, and administrative tasks) at TTK UAS, to

identify challenges and support their resolution

through targeted training initiatives. Accordingly, the

following research questions were formulated:

▪ What is the level of awareness and familiarity

with GenAI tools among faculty at TTK UAS?

▪ In which work fields are these tools most used?

▪ What are the perceived potentials and concerns

regarding GenAI across different fields at TTK

UAS?

▪ What AI-related training and guidance do

faculty members need to ensure that the use of

AI tools is both effective and aligned with

academic ethical principles?

On one hand, by mapping the opportunities and

risks associated with the use of AI-based applications

in the local higher education landscape and

Petjärv, B., Retsnoi, V., Uukkivi, A., Vilms, M., Saﬁulina, E. and Labanova, O.

Practices, Challenges, and Training Needs of Faculty in Terms of Generative AI.

DOI: 10.5220/0013287600003932

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 1, pages 83-91

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

comparing these findings with previous research, the

survey results will facilitate more focused and precise

planning of faculty training needs on site. On the

other hand, by contributing to the collection of data

on AI trends in education, this approach will enable

cross-border insights and support the design of a

common value space, providing support and training

on the topics that present the greatest challenges at

this time, where technological changes are rapid, but

adaptation to these changes takes time and requires a

systematic approach.

Therefore, before conducting the survey, the

authors analysed previously published literature, with

the results appearing in a separate article titled

“Integrating Artificial Intelligence in Higher

Education: A Literature Review of Current Trends,

Challenges, and Future Directions” (Safiulina et al.,

2024). The key findings from this article have been

integrated with the results of the current literature

review, focusing on more recent studies.

2 LITERATURE REVIEW

Growing research (Kung et al, 2023; Lund, 2023; Lee

et al., 2024; Lepik, 2024; Peres et al, 2023; Rahman

& Watanobe, 2023; Safiulina et al., 2024; Strzelecki,

2023; Ward et al, 2024) has explored faculty

perceptions of generative AI tools in higher

education, highlighting the expanding role of AI in

various academic tasks beyond teaching, such as

research activities, project-based work, and

administrative duties like drafting emails and

providing feedback. These tools support personalized

learning, enhance teaching materials, provide

research input, and assist with drafting responses to

student inquiries.

In addition to these benefits, recent literature

(Safiulina et al., 2024) identifies the role of AI in

personalized and adaptive learning as one of the most

transformative applications in higher education. AI

enables tailored educational experiences by adjusting

content to individual student needs and learning

styles, improving engagement and learning

outcomes. This finding aligns with the growing

recognition of AI's ability to offer personalized

feedback and improve the efficiency of educational

delivery (Ward et al., 2024; Lee et al., 2024).

Moreover, AI's impact on assessment processes

has been increasingly noted. Automated grading

systems and AI-enabled exams offer consistent and

timely evaluations, streamlining assessment tasks.

However, alongside these efficiencies, challenges

such as ethical concerns around fairness and bias in

AI-driven assessments remain significant (Neumann,

Rauschenberger & Schön, 2023). Safiulina et al.

(2024) also highlight this concern, emphasizing the

need for clear governance frameworks to ensure

equitable use of AI in assessment and teaching.

However, significant challenges have also been

identified regarding using AI-based text generators

like ChatGPT and other generative AI tools in

education. These include concerns about reliability,

the use of biased data (Obaid & Yaseen, 2023), the

generation of inaccurate or fabricated content,

including fictitious citations (Rahman & Watanobe,

2023), and the potential for over-reliance on AI,

which may negatively affect students’ critical

thinking and problem-solving skills (Neumann,

Rauschenberger, & Schön, 2023). These concerns

align with broader discussions on AI ethics,

emphasizing the importance of transparency and

fairness in using AI systems for educational purposes

(Holmes & Miao, 2023; Yusuf, Pervin, & Román-

González, 2024).

The ethical concerns surrounding AI use are not

limited to bias. Privacy, data security, and the

responsible use of AI are prominent in the literature

(Safiulina et al., 2024). For instance, AI systems that

collect and analyse student data raise privacy and

security concerns, particularly regarding

unauthorized access to sensitive information. Higher

education institutions must address these concerns to

ensure the responsible adoption of AI tools (Safiulina

et al., 2024; Lepik, 2024).

The response from higher education institutions

(HEIs) has varied, ranging from those enforcing strict

limitations on the use of ChatGPT (Rahman &

Watanobe, 2023) to those developing guidelines on

the ethical and responsible use of AI tools (Neumann,

Rauschenberger, & Schön, 2023). Based on the

literature review for the current study, it can be

broadly stated that there is a relatively high level of

awareness and familiarity with generative AI tools

among faculty members across a wide geographical

range and multicultural backgrounds. However,

common to these studies is the finding that faculty

require support and training to address concerns

related to academic integrity (Lee et al., 2024; Ward

et al., 2024; Yusuf, Pervin, & Román-González,

2024). As Safiulina et al. (2024) suggest, there is an

urgent need for AI literacy among both educators and

students, as well as comprehensive institutional

strategies that integrate AI into professional

development programs to enhance AI competency.

As Chiu (2024) suggests, rather than issuing strict

recommendations, institutions should develop

guidelines and policies that emphasize the

CSEDU 2025 - 17th International Conference on Computer Supported Education

competencies needed for the future workforce,

supported by in-class and hands-on activities. To

fully comprehend the impact of generative AI on

assessment, AI and generative AI should be

integrated into teacher professional development

programs within universities. D. Ward and his

research group emphasize that, given the

complexities of generative AI, faculty require time

and resources to effectively learn and adapt their

classes to help students engage with AI ethically and

critically; therefore, universities should creatively

develop AI-powered learning assistants, adaptive

learning systems, and faculty support tools while

focusing on inclusiveness, transparency, privacy, and

safety, all with a steadfast commitment to enhancing

human interaction and improving the quality of

teaching and learning (Ward et al., 2024). The most

comprehensive treatment of this topic has been

provided in the Guidance for generative AI in

education and research (Holmes & Miao, 2023).

Additionally, Lepik (2024) indicates that instructors

prefer regular, specific, and field-relevant training

formats to address ongoing challenges, further

underscoring the need for structured institutional

support.

3 METHODOLOGIES

3.1 Sample

This study, conducted in June 2024, utilized a

quantitative approach to assess the perceptions of

academic staff regarding the use of GenAI in

educational settings. The survey was partly based on

a similar staff survey conducted in February 2024 at

the University of Tartu, Estonia's largest university

(ranked #358 in the QS World University Rankings),

which focused solely on the use of text-generating

bots in teaching.

The data for the current study were collected

through an online survey administered via Google

Forms, ensuring ease of access and participation for

all employees. The survey was designed to be self-

administered, allowing respondents to fill out the

questionnaire independently. This approach

facilitated anonymity and encouraged honest

responses, which is critical for obtaining accurate and

reliable data.

In the 2023/24 academic year, the total number of

TTK UAS academic staff was 237. The study group

consisted of 81 lecturers who responded to the survey.

The response rate in the population was approximately

34.18%. Participants were ranked as follows by

discipline: engineering (71.60%) and social sciences

and humanities (28.40%); by teaching experience: up

to 3 years (20.99%), 4–6 years (9.88%), 7–9 years

(11.11%), and 10+ years (58.02%); and by AI training

experience: has previously participated in AI training

(51.85%) and has not previously participated in AI

training (48.15%).

3.2 Data Collection

The survey comprises five distinct subscales, each

focusing on different aspects of generative AI usage

among lecturers. These include the Purpose of Using

Generative AI, Capabilities of Using Generative AI

Tools in Teaching, Risks Associated with Using

Generative AI Tools in Teaching, Training Needs in

AI, and Workload Dynamics of Academic Staff. The

responses for each of the items are summed to give a

total score.

The ‘Purpose of Using Generative AI’ subscale

(Cronbach’s alpha 0.77) is an 8-item scale used to

measure the extent to which lecturers use AI in the

preparation and conduct of teaching, research, writing

documentation, answering emails, etc. The total score

ranges from 2 to 27, with a higher score indicating

more frequent use of generative AI.

The ‘Capabilities of Using Generative AI tools in

Teaching’ subscale (Cronbach’s alpha 0.84) is an 8-

item scale used to measure how useful generative AI

can be in teaching from a lecturer's perspective. The

total score ranges from 13 to 38, with a higher score

indicating that the lecturers see more opportunities for

using generative AI tools in teaching.

The ‘Risks Associated with Using Generative AI

Tools in Teaching’ subscale (Cronbach’s alpha 0.88)

is a 9-item scale used to measure how risky generative

AI can be in teaching from a lecturer's perspective.

The total score ranges from 14 to 45, with a higher

score indicating that the lecturers see more risks

associated with using AI tools in teaching.

The ‘Training Needs in AI’ subscale (Cronbach’s

alpha 0.82) is an 11-item scale used to assess which

topics should be covered in future AI training for

academic staff. The total score ranges from 31 to 55,

with a higher score indicating that lecturers need

more diverse training on how to effectively apply

generative AI tools in teaching.

The ‘Workload Dynamic of Academic Staff’

subscale is a 6-item scale (Cronbach’s alpha 0.89)

used to assess how the use of generative AI has

affected lecturers' workload. The total score ranges

from 0 to 18, with a higher score indicating that

lecturers perceive an increase in workload due to the

use of generative AI.

Practices, Challenges, and Training Needs of Faculty in Terms of Generative AI

3.3 Analysis

Firstly, the descriptive statistics for the relevant scale

variables were calculated. Then, the relationships

between these scale variables were investigated.

Finally, it was examined whether these variables

differed significantly according to independent

variables such as institution, teaching discipline

(engineering or social sciences/humanities), teaching

experience, and whether the individuals were AI

trained or not.

Data were analysed using MS Excel and the

statistical software R. Correlations between variables

were examined using Spearman’s rho coefficient.

Differences between groups were assessed using the

Wilcoxon rank-sum test and the Kruskal-Wallis rank-

sum test. Significance was set at a minimum level of

0.05, with other significance levels (0.01 and 0.001)

also reported.

4 RESULTS

The analysis of faculty awareness and usage of

GenAI tools at TTK UAS revealed varied levels of

familiarity, tool preference, and impact on academic

workload. The findings indicated that most faculty

members are acquainted with generative AI tools,

particularly text bots, with 82.72% of respondents

reporting usage in their work. This preference is

followed by image generators (29.63%), audio

generators (8.64%), and video generators (7.41%).

Notably, 16.05% of respondents do not utilize any

GenAI tools in their professional activities.

The previous question was expanded on how the

use of these tools has affected faculty’s workload?

The ‘Workload Dynamic of Academic Staff’ scale

consists of six items measured on the following scale:

0 – ‘Not relevant to me,’ 1 – ‘Decreased,’ 2 –

‘Remained the same,’ and 3 – ‘Increased.’

Descriptive statistics for the relevant scale variables

were calculated and the in-depth examination

revealed that, in most cases, AI adoption led to faculty

workloads either remaining consistent or showing a

slight shift, either increasing or decreasing. It is

essential to specify that the responses reflect

perceived values when addressing how the use of

these tools has affected faculty's workload. Faculty

did not perceive a change in workload in Student

Support and Feedback (M=2.00, SD=0.58) and in

Research (M=1.82, SD=0.56). The workload

remained the same or slightly decreased in

Preparation of Teaching Materials (M=1.75,

SD=0.61), in Project Work (M=1.65, SD=0.57) and

in Administrative Duties - such as email management

and routine communication (M=1.71, SD=0.54). The

workload tended to remain stable too or showing

slight increase in Assessment of Student Work

(M=2.23, SD=0.56). Note that the workload score

tends to be higher when the lecturer's field of activity

is broader, rather than reflecting a genuine increase in

workload.

The ‘Purpose of Using Generative AI’ scale

consisted of eight items measured on the following

scale: 0 – ‘Not Relevant to Me’, 1 – ‘No, I Don't Plan

to Start’, 2 – ‘No, But I Plan to Start’, 3 – ‘Yes,

Occasionally’, and 4 – ‘Yes, Regularly’. Descriptive

statistics for the relevant scale variables were

calculated and the most frequent usage of generative

AI tools lies in preparing teaching materials (M=2.68,

SD=0.84). Generative AI tools for conducting

teaching activities in exercises and practical sessions

(M=2.32, SD=0.98), for research-related tasks

(M=2.31, SD=0.94), and for writing project

documentation (M=2.29, SD=0.95) were viewed as

promising but not yet fully adopted. The application

of AI in student-related tasks, particularly for non-

graded assignments (M=1.99, SD=0.87) and graded

evaluations (M=1.89, SD=0.90), remained relatively

low, as well the same applies also for AI assistance in

managing email correspondence (M=1.89 (SD=1.03).

Generative AI tools were the least commonly used for

guiding students and providing feedback (M=1.64,

SD=0.84).

In this survey, there were two additional questions

about how lecturers regulate the use of AI in practical

work with students and the reasons for refraining

from using AI in teaching. For these questions,

multiple-choice answers were permitted. 45.7% of

respondents allow students to use text bots freely

without altering teaching methods. 35.8% incorporate

text bots while educating students on associated risks

and opportunities. 17.3% have adjusted their teaching

methods to integrate AI tools. A small portion (4.9%)

prohibit AI use and enforce this restriction.

Common reasons for avoiding AI tools in

teaching centred on concerns about reliability,

guidance, and educational impact: 40.7% expressed

doubts about the reliability of AI-generated

information. 35.8% cited a lack of guidance on

responsible and ethical AI usage. 33.3% were

concerned about AI's impact on student knowledge.

Other notable reasons included AI errors (25.9%),

incompatibility with subjects taught (23.5%), and

difficulty in effective AI use (23.5%).

CSEDU 2025 - 17th International Conference on Computer Supported Education

Table 1: Means (M) and standard deviations (SD) for the ‘Capabilities of Using Generative AI Tools in Teaching’ items

(n=81).

Capabilities of Using Generative AI Tools in Teaching M SD Conclusion

Enables the quick creation of personalized learning materials 3.30 0.93 Neutral,

dissension

Makes teaching more practical and closer to real life by supporting the use of

active learning methods

3.12 1.08 Neutral,

dissension

Enables automated grading and feedback 3.26 0.95 Neutral,

dissension

Encourages experimentation with new pedagogical approaches and promotes

creativit

3.69 1.03 Rather Agree

Supports the development of more systematic and analytical thinking 2.85 1.12 Neutral,

dissension

Develops skills in students that will be required in the future labour market

(including expanding current career opportunities and enhancing their ability to

succee

in the labo

market

)

3.64 1.02 Rather Agree

Does not provide extra opportunities to enrich the study (the reverse scale is used) 3.80 0.97 Rather Disagree

Complements traditional teaching methods 3.88 0.97 Rather Agree

Table 2: Means (M) and standard deviations (SD) for the ‘Risks Associated with Using Generative AI Tools in Teaching’

items (n=81).

Risks Associated with Using Generative AI Tools in Teaching M SD Conclusion

Raises the risk of academic fraud because AI-generated content is difficult to

detect

4,20 0,90 Rather Agree

Produces unreliable content, which in turn threatens academic integrity and

distorts the worldview

4,01 0,93 Rather Agree

Does not support the development of students' critical thinking and makes the

learnin

rocess "co

aste"

ase

3,86 1,16 Rather Agree

Causes security risks related to the safety of users of systems that manage AI tools

to the unfai

use of

ersonal data

3,63 0,98 Rather Agree

Blurs the boundaries and principles of ethical and unethical academic behaviour 3,81 1,00 Rather Agree

It has a negative impact on the development of students' mental and emotional

health

3,00 0,92 Neutral, dissension

Reduces the component of creativity and originality in students' work 3,43 1,22 Neutral, dissension

Makes it difficult to treat students equally 3,63 1,17 Rather Agree

There are no significant risks associated with using AI-based tools in education

(

the reverse scale is used

)

3,79 1,10 Rather Disagree

The ‘Capabilities of Using Generative AI Tools in

Teaching’ and ‘Risks Associated with Using

Generative AI Tools in Teaching’ scales each

consisted of items measured on a 5-point Likert scale:

1 – ‘Strongly Disagree’, 2 – ‘Disagree’, 3 – ‘Neutral’,

4 – ‘Agree’, and 5 – ‘Strongly Agree.’ Descriptive

statistics for the relevant scale variables were

calculated and are presented in Tables 1 and 2.

Among the open responses, it was highlighted that

integrating AI provides the opportunity to break

exercises down into smaller parts or combine them

into a whole, which supports learning. It also helps to

offer different examples from practical life for the

solutions to a single task. AI-based technology has

provided new possibilities for doing things differently

and for experiencing both learning and teaching in

new ways. In the open responses, a direct threat to the

future of engineering was even emphasized,

suggesting that allowing AI-based solutions could

create a springboard for charlatans in the field, posing

a danger to society.

To evaluate the overall impact of generative AI on

higher education, a 5-point Likert scale was used: 1 –

‘Very Negative,’ 2 – ‘Rather Negative,’ 3 – ‘Neutral,’

4 – ‘Rather Positive,’ and 5 – ‘Very Positive. 61.73%

of participants rated it as rather or very positive,

20.99% as neutral, and 17.28% as rather or very

negative. In assessing the overall impact, speed-

related aspects (more information, faster) were the

most frequently mentioned in the comments. As a

counterpoint to the ability to process information

more quickly, open responses also highlighted that

Practices, Challenges, and Training Needs of Faculty in Terms of Generative AI

the issue is not just about the best technical solution

but about what it should enable in one’s ongoing

work—something that isn’t always clear, with

significant time potentially spent on adjusting

workflows (meta-work). Overall, it was noted that

achieving a positive impact requires proper guidance

and the parallel development of supportive skills such

as critical thinking and creativity; otherwise, the role

of independent thinking may diminish. Descriptive

statistics and correlations of the relevant scale

variables and AI overall impact rate were calculated

for the total sample (Table 3 and 4). The results

indicate, similarly to the previous findings, that the

participants rated the overall impact of generative AI

tools on higher education as rather positive on a 5-

point Likert scale (M=3.49, SD=0.88).

There was also a moderate positive correlation

between the ‘Capabilities of Using Generative AI’

and ‘Training Needs in AI’ (ρ=0.30, p=0.0058), as

well as the ‘Overall Impact Rate of AI’ (ρ=0.61,

p<0.001). A Spearman’s rho correlation analysis

found a moderate negative correlation between ‘Risks

Associated with Generative AI’ and ‘Purpose of

Using Generative AI’ (ρ=-0.32, p=0.0042),

‘Capabilities of Using Generative AI’ (ρ=-0.495,

p<0.001), and ‘Overall Impact Rate of AI’ (ρ=-0.52,

p<0.001).

There was no significant correlation between

‘Risks Associated with Generative AI’ and ‘Training

Needs in AI.’ Similarly, a weak positive correlation

was found between ‘Training Needs in AI’ and

‘Overall Impact Rate of AI’ (ρ=0.26, p=0.0175).

However, no significant correlation was found

between the ‘Workload Dynamics of Academic Staff’

and other relevant variables.

The Wilcoxon rank-sum test and the Kruskal-

Wallis rank-sum test were used to assess statistically

significant differences between groups based on

institution, teaching discipline (engineering or social

sciences/humanities), teaching experience, and

whether the individuals were AI trained or not. It was

found that there were no statistically significant

differences between the groups based on institution

and teaching experience with respect to all relevant

variables. Additionally, the Wilcoxon rank sum test

revealed that the total scores for ‘Risks Associated

with Generative AI’ differ significantly according to

the lecturers’ teaching discipline (W=891,

p=0.01905). Specifically, the test results indicate that

the total scores for ‘Risks Associated with Generative

AI’ were higher for lecturers who teach engineering

disciplines than for those who teach social sciences or

humanities.

Table 3: Descriptive Statistics of the relevant scale variables and the overall impact rate of AI (n=81).

Variable M SD MIN 25% 50% 75% MAX Skewness Kurtosis

Purpose of Using AI 13.53 5.86 0 10 13 17 27 -0.17 0.24

Capabilities of Using AI 27.54 5.51 13 24 28 32 38 -0.36 0.15

Risks Associated with AI 33.37 6.80 14 30 34 38 45 -0.69 0.47

Training Needs in AI 47.54 5.67 31 44 48 52 55 -0.82 0.42

Workload Dynamic of Acad. Staff 6.57 4.88 0 2 7 11 18 0.03 -1.26

Overall Impact Rate of AI 3.49 0.88 1 3 4 4 5 -0.71 -0.15

Table 4: Spearman’s rho coefficients for the relevant scale variables and the overall impact rate of AI (n=81).

Purpose of

Using AI

Capabilities

of Using AI

Risks

Associated

with AI

Training

Needs in

Workload

Dynamic of

Academic Staff

Overall

Impact

Rate of AI

Purpose of Using AI 1

Capabilities of Using AI 0.45*** 1

Risks Associated with AI -0.32** -0.495*** 1

Training Needs in AI 0.29** 0.30** 0.07 1

Workload Dynamic of

Academic Staff

0.19 0.08 0.001 -0.04 1

Overall Impact Rate of AI 0.49*** 0.61*** -0.52*** 0.26* 0.03 1

***p<0.001 **p<0.01 *p<0.05

CSEDU 2025 - 17th International Conference on Computer Supported Education

Table 5: Means (M) and standard deviations (SD) for the ‘Training Needs in AI’ items (n=81).

Training for academic staff should focus on the following topics M SD Conclusion

Practical recommendations for using text bots in planning and organizing teaching,

including evaluation an

feedbac

4,49 0,63 Rather Agree

Text bots (including those that generate images, videos, and audio) and academic

frau

4,46 0,78 Rather Agree

Learning methods in the age of AI 4,38 0,66 Rather Agree

The future of the labor market and the skills required by graduates 4,21 0,89 Rather Agree

Enhancing the effectiveness of teaching staff using AI tools 4,48 0,63 Rather Agree

Legal and ethical aspects of data protection and the use of text, image, video, and

audio generators

4,43 0,79 Rather Agree

Equal treatment of students in using generative AI tools 4,15 1,00 Rather Agree

The risks associated with the use of generative AI tools for humanity (including

ative environmental im

acts

)

3,98 1,12 Rather Agree

The use of text bots in scientific research 4,26 0,92 Rather Agree

The use of text bots in project work 4,27 0,84 Rather Agree

There is no need for additional training (the reverse scale is used) 4,43 1,01 Rather Disagree

The Wilcoxon rank sum test also found that the

total scores for ‘Risks Associated with Generative

AI’ (W=1139.5, p=0.00245) and ‘Capabilities of

Using Generative AI’ (W=606, p=0.04393) differ

significantly between lecturers who had previously

attended AI training and those who had not.

Specifically, lecturers trained in AI perceive fewer

risks and more opportunities in using generative AI

in teaching compared to those who are untrained.

Finally, we collected information on which

topics faculty believe should receive the most

attention in training. The ‘Training Needs in AI’

scale consists of items measured on a 5-point Likert

scale: 1 – ‘Strongly Disagree,’ 2 – ‘Disagree,’ 3 –

‘Neutral,’ 4 – ‘Agree,’ and 5 – ‘Strongly Agree.’

Descriptive statistics for the relevant scale variables

were calculated and are presented in Table 5.

5 DISCUSSIONS

The findings of this study provide detailed insights

into the use and perceptions of GenAI among

faculty members at TTK UAS, which align closely

with trends observed in the broader academic

literature. The existing literature emphasizes the

potential for AI tools to transform higher education,

particularly in areas such as personalized learning

and automated assessments (Ward et al., 2024;

Safiulina et al., 2024). Similarly, the findings of this

study indicate that faculty members at TTK UAS

recognize the potential of GenAI tools in preparing

teaching materials (M=2.00, SD=0.58) and

conducting research (M=1.82, SD=0.56).

The literature emphasizes the importance of

addressing ethical concerns and ensuring the

transparency of AI systems (Holmes & Miao, 2023;

Neumann, Rauschenberger, & Schön, 2023). The

current study reinforces these findings by

identifying reliability (40.7%) and ethical concerns

(35.8%) as significant barriers to the adoption of

GenAI tools. The survey results also highlight

practical challenges, such as a lack of training and

guidance, which align with previous studies (Chiu

et al., 2023; Lepik, 2024; Safiulina et al., 2024;

Ward et al., 2024) advocating for structured

institutional support and professional development

programs.

While the literature (Chiu et al., 2023; Lee et al.,

2024; Neumann et al., 2023; Ward et al., 2024)

underscores the efficiency gains associated with AI

in educational settings, the workload dynamics

observed in this study present a more complex

picture. Although AI tools are seen as beneficial in

streamlining certain tasks, their impact on workload

appears to vary across different academic activities.

The observed slight increase in workload related to

the Assessment of Student Work, as reported in this

study, may reflect the additional effort required to

evaluate AI-generated student submissions. In this

context, addressing workload concerns underscores

the importance of employing appropriate

pedagogical frameworks and implementing more

nuanced strategies for integrating AI tools, ensuring

that faculty workloads are optimized without

introducing additional burdens.

One of the key contributions of this study is the

identification of discipline-specific differences in

perceptions of GenAI risks and opportunities.

Practices, Challenges, and Training Needs of Faculty in Terms of Generative AI

Engineering faculty members, for instance, reported

higher levels of concern regarding AI risks

compared to their counterparts in social sciences

and humanities. This aligns with previous research

suggesting that the perceived applicability and risks

of AI tools can vary widely depending on the

disciplinary context (Rahman & Watanobe, 2023).

The findings also reveal significant correlations

between faculty perceptions of GenAI capabilities,

risks, and training needs. Faculty members with

prior AI training were more likely to perceive

opportunities and fewer risks associated with

GenAI. This underscores the critical role of training

in shaping positive attitudes and facilitating the

effective integration of AI tools in teaching and

research.

Overall, this study extends the existing literature

by providing localized insights into the challenges

and opportunities associated with GenAI in higher

education. Challenges, particularly in guiding

students and providing personalized learning

experiences, can be addressed through well-

designed training programs that support faculty, as

the literature suggests that GenAI tools can play a

critical role in assisting vulnerable student groups,

including those with learning disabilities (Lee et al.,

2024). While 45.7% of TTK UAS faculty allow

students to use AI without adapting teaching

practices—aligning with findings from the

University of Tartu study, where 22.03% reported

similar permissiveness—this may reflect a lack of

awareness or readiness to address associated risks

and their underlying causes. In this context,

practical, goal-oriented training on integrating

GenAI tools into teaching, in alignment with

academic and ethical values, is essential, as outright

prohibition is neither practical nor sustainable.

The study also reveals that generative AI tools

are moderately utilized in research-related tasks and

project documentation, yet faculty express a desire

for additional training in these areas, indicating

untapped potential for streamlining such activities.

Faculty at TTK UAS have expressed strong interest

in actionable, discipline-specific guidance,

emphasizing the need for institutional efforts to

prioritize these areas to fully harness AI’s potential

while mitigating risks and promoting equitable

educational practices.

These findings underscore the importance of

developing tailored educational programs and

institutional policies that address both the ethical

and practical dimensions of AI integration.

6 CONCLUSIONS

This study contributes to the growing body of

research on the integration of generative AI in

higher education by providing empirical evidence

on faculty perceptions, usage patterns, and training

needs at TTK UAS. The findings highlight that

while GenAI tools have significant potential to

enhance teaching, research, and administrative

tasks, their effective implementation requires

addressing key challenges related to ethical

concerns and AI literacy.

The study underscores the need for targeted

training programs to equip faculty with the skills

and knowledge needed to effectively use AI tools

while adhering to ethical guidelines. By

emphasizing discipline-specific needs, institutions

can ensure that AI tools are integrated in ways that

enhance educational quality without compromising

academic integrity.

The identification of correlations between

training, perceptions of risks, and opportunities

suggests that increasing access to AI training could

play a pivotal role in overcoming barriers to

adoption. Additionally, the underutilization of AI in

student-focused tasks indicates a need for further

exploration of how these tools can enhance student

engagement and learning outcomes.

Future research should examine the longitudinal

impact of AI integration on both faculty workload

and student learning outcomes. Additionally, there

is a need for comparative studies across institutions

to identify best practices and develop standardized

frameworks for the ethical and effective use of AI

in higher education. Such efforts help higher

education institutions manage the challenges of

adopting AI, leading to a more inclusive and

innovative learning environment.

REFERENCES

Castelli, M., & Manzoni, L. (2022). Generative models in

artificial intelligence and their applications. Applied

Sciences, 12(9), 4127.

Chiu, T. K., Xia, Q., Zhou, X., Chai, C. S., & Cheng, M.

(2023). Systematic literature review on opportunities,

challenges, and future research recommendations of

artificial intelligence in education. Computers and

Education: Artificial Intelligence, 4, 100118.

Holmes, W., & Miao, F. (2023). Guidance for generative

AI in education and research. UNESCO Publishing.

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

Leon, L., Elepaño, C., ... & Tseng, V. (2023).

Performance of ChatGPT on USMLE: potential for

CSEDU 2025 - 17th International Conference on Computer Supported Education

AI-assisted medical education using large language

models. PLoS digital health, 2(2), e0000198.

Lee, D., Arnold, M., Srivastava, A., Plastow, K., Strelan,

P., Ploeckl, F., ... & Palmer, E. (2024). The impact of

generative AI on higher education learning and

teaching: A study of educators’ perspectives.

Computers and Education: Artificial Intelligence, 6,

100221.

Lepik, K. (2024, May 17). Tartu Ülikooli õppejõudude

praktikad ja probleemid seoses tekstirobotite

kasutamisega õppetöös [Presentation slides]. E-

kursuse kvaliteedimärgi lõpuseminar, Tallinn.

Lund, B. D., Wang, T., Mannuru, N. R., Nie, B., Shimray,

S., & Wang, Z. (2023). ChatGPT and a new academic

reality: Artificial Intelligence‐written research papers

and the ethics of the large language models in

scholarly publishing. Journal of the Association for

Information Science and Technology, 74(5), 570-581.

Neumann, M., Rauschenberger, M., & Schön, E. M.

(2023, May). “We need to talk about ChatGPT”: The

future of AI and higher education. In 2023 IEEE/ACM

5th International Workshop on Software Engineering

Education for the Next Generation (SEENG) (pp. 29-

32). IEEE.

Obaid, O. I., Ali, A. H., & Yaseen, M. G. (2023). Impact

of Chat GPT on Scientific Research: Opportunities,

Risks, Limitations, and Ethical Issues. Iraqi Journal

for Computer Science and Mathematics, 4(4), 13-17.

Peres, R., Schreier, M., Schweidel, D., & Sorescu, A.

(2023). On ChatGPT and beyond: How generative

artificial intelligence may affect research, teaching,

and practice. International Journal of Research in

Marketing, 40(2), 269-275.

Rahman, M. M., & Watanobe, Y. (2023). ChatGPT for

education and research: Opportunities, threats, and

strategies. Applied Sciences, 13(9), 5783.

Safiulina, E., Labanova, O., Uukkivi, A., Petjärv, B., &

Vilms, M. (2024). Integrating Artificial Intelligence

in Higher Education: A Literature Review of Current

Trends, Challenges, and Future Directions.

ICERI2024 Proceedings (In print)

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, 1-14.

Ward, D., Loshbaugh, H. G., Gibbs, A. L., Henkel, T.,

Siering, G., Williamson, J., & Kayser, M. (2024).

How Universities Can Move Forward With

Generative AI in Teaching and Learning. Change:

The Magazine of Higher Learning, 56(1), 47-54.

Yusuf, A., Pervin, N., & Román-González, M. (2024).

Generative AI and the future of higher education: a

threat to academic integrity or reformation? Evidence

from multicultural perspectives. International

Journal of Educational Technology in Higher

Education, 21(1), 21.

Practices, Challenges, and Training Needs of Faculty in Terms of Generative AI