Generative AI for Productivity in Industry and Education

Ferenc H

ejja

1,3 a

, Tam

as Bart

3 b

, Roy Dakroub

3 c

and Gergely Kocsis

2 d

University of Debrecen, Doctoral School of Informatics, Hungary

University of Debrecen, Faculty of Informatics, Department of Informatics Systems and Networks, Hungary

EPAM Systems, Hungary

Keywords:

Generative Artiﬁcial Intelligance (GenAI), Large Language Models (LLM), Industry, Education, Productivity.

Abstract:

Generative AI tools are the cutting edge solutions of complex AI related problems. While investigating state-

of-the-art results related to the effect of GenAI in the literature, one can note that the trends most likely lead to

the expectation of a positive effect on the middle and long run. Based on these ﬁndings we deﬁne 4 productivity

gain related hypotheses that we study using two types of methodologies. Namely we perform a survey research

related to university-industry collaboration and quantitative studies mainly based on industrial productivity

metrics. We have partnered with a major IT services provider - EPAM Systems - to be able to track, validate

and analyze the key productivity metrics of software development projects, with and without using GenAI

tools. This evaluation is being performed on various stages of the Software Development Lifecycle (SDLC)

and on several project roles. Our goal is to measure the productivity increase provided by GenAI tools.

Although this research has just started recently, considering that the area has extremely high attention we

present some initial ﬁndings.

1 INTRODUCTION

According to a recent note published by IMF ”Al-

most 40 percent of global employment is exposed to

AI” (Cazzaniga et al., 2024). Even though the ab-

solute impact of this exposure is not uniform, indus-

try and especially IT ecosystem has to stay up to date

with the emerging changes implied by AI and more

precisely LLM models. The same note emphasizes

that ”Workers with a college education have histori-

cally shown a greater ability to transition into what

are now jobs with high AI-complementarity poten-

tial.” underlining the well-known fact that the key of

making Industries able to adopt to these changes is

education (Cazzaniga et al., 2024).

Research by Cecilia Ka Yuk Chan, Wenjie Hu,

and Faming Wang’s team unveils that students’ per-

ceptions of GenAI signiﬁcantly inﬂuence their learn-

ing outcomes, and that a carefully planned AI educa-

tion policy, like the proposed AI Ecological Education

Policy Framework, can help manage AI integration in

https://orcid.org/0009-0008-5770-5577

https://orcid.org/0009-0005-3384-1919

https://orcid.org/0000-0002-7767-1107

https://orcid.org/0000-0003-0018-4201

university settings, align actions with their policy, im-

prove AI literacy and thereby, prepare students for an

AI-driven future. (Chan, 2023; Chan and Hu, 2023)

(Wang et al., 2023).

Not surprisingly one of the areas of industry that

is the most affected by the revolution of Generative

AI is the IT industry. While the last revolution of

this ﬁeld was the widespread application of agile soft-

ware development methodologies about a decade ago,

it seems that nowadays we are at the rise of another

revolution. Even though the exact productivity gain of

using AI at different phases of the Software Develop-

ment Life Cycle (SDLC) is not known, and there are

even voices saying that in some cases the use of AI

can even hinder the production, it seems that accord-

ing to the common voices the use of LLM Generative

AI (GenAI) can improve productivity. The questions

we pose in this current research aim to ﬁnd some ex-

act measures of this productivity gain so that one can

decide that according to the recent state of the tech-

nology at what phases of SDLC does it worth to apply

the new tools.

In the meantime we try to apply our results from

three different points of view. i.) The ﬁrst and most

important aspect is of course the industrial one since

this is the ﬁeld that can beneﬁt the most in the short

128

Héjja, F., Bartók, T., Dakroub, R. and Kocsis, G.

Generative AI for Productivity in Industry and Education.

DOI: 10.5220/0012736200003708

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

In Proceedings of the 9th International Conference on Complexity, Future Information Systems and Risk (COMPLEXIS 2024), pages 128-135

ISBN: 978-989-758-698-9; ISSN: 2184-5034

run from applying AI. ii.) The second aspect is from

the perspective of industry management. It may be

a question how AI can improve productivity on high

level management of industries. This connection ap-

pears to be less trivial. iii.) The third aspect is the

side of education. More precisely university and col-

lege education. The main questions here is if these

actors can effectively prepare students to the appli-

cation of GenAI in their work and even it would be

good to know what effects GenAI has on the learn-

ing experience itself. It is also an open questionn

how the high education can integrate the industrial re-

sults and tasks into their programme as new special-

izations (e.g. How to prepare students to be suitable

for new job types like prompt engineering or GenAI

aided software development?).

Since these three ﬁelds and questions cover a re-

ally broad area of research we aim to focus our stud-

ies the exact case IT industry and related studies in

the early phase of our work.

In this paper the next section presents our ﬁnd-

ings based on reviewing related state-of-the-art pub-

lications. After that we pose our hypotheses we aim

to answer in our research, while in section 4 we show

the two forms of studies we would like to use we also

present our ﬁrst ﬁndings that may prove our hypothe-

ses. The paper closes with a short discussion.

2 LITERATURE REVIEW

2.1 Generative AI in the IT Industry

Although the application of Generative AI in indus-

trial processes does not have a broad scientiﬁc liter-

ature yet, leading IT companies has started to pub-

lish their related ﬁndings in the industrial environ-

ment (including white papers, technical reports and

business journals). These publications predict that we

are before fundational productivity changes in the re-

lated ﬁelds of industry.

According to EPAM’s report, ”A Call to Ac-

tion for Generative AI” (Burkitt et al., 2023), 80%

of the workforce could have at least 10% of their

tasks affected, 19% of the workforce may see at least

50% of their tasks impacted, 300 million full-time

jobs could potentially be automated globally, Genera-

tive AI could eventually increase annual global GDP

by 7%, Productivity gains for a range of tasks and

processes may be greater than 50%, The combined

impact of productivity gains and revenue growth

may increase the enterprise value of successful early

adopters by up to 20%+ (see also (Eloundou et al.,

2023) and (Hatzius et al., 2023)).

A study conducted by William Harding and

Matthew Kloster from GitClear suggests that AI pro-

gramming assistants such as GitHub Copilot could

decrease code quality and increase redundancy. The

study reveals AI tools are proﬁcient at adding new

code but fail to update, delete, or move existing ones,

resulting in an alarming increase in code churn and re-

dundancy. Additionally, concerns about AI-generated

code’s security have also surfaced in other studies.

Despite these concerns, its positive impact on pro-

ductivity is acknowledged, contingent upon task com-

plexity and developer skill. Nevertheless, a consensus

seems to indicate humans are irreplaceable in coding,

as AI tools are still error-prone. (Harding and Kloster,

2024)

In recent research conducted by Thomas Dohmke,

Marco Iansiti, and Greg Richards, generative AI, in-

cluding GitHub Copilot, has been found to signif-

icantly increase developer productivity. The tool

was shown to help developers implement solutions

faster, leading to improved productivity and satisfac-

tion. GitHub Copilot’s impact only grows over time,

with users accepting an average of 30% of code sug-

gestions and less experienced developers beneﬁting

the most. The researchers argue that as developers

become more proﬁcient in AI-prompting and interac-

tion, approximately 80% of code will be AI-written

in the future — a trend which could democratize soft-

ware development and boost developers’ innovative

potential. Like previous groundbreaking technolo-

gies, generative AI may lead to new business mod-

els and a shift towards higher-order work. (Dohmke

et al., 2023)

In this study by Alok Mishra and Yehia Ibrahim

Alzoubi, Agile and Waterfall methodologies were

compared and analyzed for software development.

The researchers discovered that both methodologies

have their strengths: Agile for its ﬂexibility and Wa-

terfall for its stability. They concluded that there is

no one-size-ﬁts-all approach; instead, ﬁrms may need

to use a hybrid framework combining aspects of both

Agile and Waterfall methods to meet different project

requirements. The study also suggested that future re-

search could focus on real-world applications of these

hybrid methodologies. Ultimately, the researchers ad-

vocated for ﬁrms to incorporate Agile principles into

their existing systems, especially in the digital era.

(Mishra and Alzoubi, 2023)

In his book ”Generative AI - Navigating the

Course to the Artiﬁcial General Intelligence Future”,

Martin Musiol invites readers on a journey into the

new world of generative AI and artiﬁcial general in-

telligence (AGI), arguing that we are on the precipice

of a transformative epoch in technology. He believes

Generative AI for Productivity in Industry and Education

129

that advancements in AI like ChatGPT mark a point

of no return, underscoring the vast potential of gener-

ative AI across a variety of ﬁelds. Musiol maintains

that the swift progression and adoption of generative

AI surpasses the growth arc of preceding technolo-

gies. Believing that this technology has the potential

to redeﬁne the future, he asserts that generative AI can

empower individuals to become signiﬁcantly more ef-

fective humans. Musiol concludes that mastering AI

will offer a distinct advantage in the realm of tomor-

row, endorsing AI as a tool for striving towards a bet-

ter future rather than a replacement for human intel-

lect and innovation. (Musiol, 2023)

2.2 Generative AI in Universities

In contrary to the industrial era, educational related

research has already a notable amount of scientiﬁc

surces in connection to the use of Generative AI tools

by students.

The research conducted by Cecilia Ka Yuk Chan

and Wenjie Hu reveals that students’ views on Gen-

erative AI (GenAI) technologies signiﬁcantly impact

their learning processes and results. By getting to

know students’ readiness and apprehensions about

GenAI tools, educators can incorporate such tech-

nologies into the teaching process more efﬁciently.

This step enhances educational outcomes and culti-

vates a comprehensive approach to learning. More-

over, understanding students’ perspectives helps in

assessing AI literacy, enabling the educators to iden-

tify and bridge the knowledge gaps, thereby prepar-

ing students for an imminent AI-driven future. (Chan,

2023; Chan and Hu, 2023).

Faming Wang and his team aimed to devise an

AI education policy targeted at university teaching

and learning. The AI Ecological Education Policy

Framework was proposed to manage the varied as-

pects of AI integration in university settings, divided

into three dimensions – Pedagogical, Governance,

and Operational. This structure aims to help stake-

holders better understand the implications of AI for

teaching and learning and ensure that they are aware

of their responsibilities. If this framework is adopted,

educational institutions can ensure the responsible

and ethical use of AI and augment potential beneﬁts.

Nonetheless, further study is necessary to fully under-

stand the potential beneﬁts and risks associated with

AI in academic settings. Rather than simply push-

ing for AI implementation, stakeholders must care-

fully consider which AI technologies to use, how best

to use them, and fully grasp their capabilities. (Wang

et al., 2023).

In a recent research study, Ramteja Sajja and

Ibrahim Demir developed an automated system, Vir-

tualTA, designed to answer logistical questions on on-

line course discussion boards and educational plat-

forms. The researchers aimed to enhance the qual-

ity of course content and individualized student ad-

vising as well as mitigate inequality amongst students

in terms of knowledge accessibility. The virtual assis-

tant system is also designed to ease transitions for stu-

dents changing degree programs or disciplines. While

the system was developed and tested under controlled

circumstances, the research offers valuable insights

into the potential of VirtualTA in enhancing student

learning and engagement. The researchers acknowl-

edge that while chatbots can reduce the workload for

teachers, they cannot entirely replace human interac-

tion. Future studies could explore improvements to

the AI-augmented educational assistance framework

to better cater to students’ diverse needs and complex-

ities in their academic journeys.(Sajja et al., 2023)

Jeya Amantha Kumar’s research study explores

the impact of Embodied Conversational Agents (ECs)

on project-based learning activities. The study aimed

to evaluate learning outcomes and found that ECs

positively inﬂuenced learning performance and team-

work. However, the study found no signiﬁcant dif-

ferences in the need for cognition, motivational be-

lief, creative self-efﬁcacy, and perception of learning.

The research concluded that while ECs did not signif-

icantly impact cognitive and motivational processes,

they did not create any barriers to project-based learn-

ing either. Interestingly, the study found that the in-

troduction of ECs contributed to a sense of ”team

spirit”. Going forward, further research could explore

the transformational potential of ECs as digital assis-

tants in educational settings.(Kumar, 2021)

While the above results clearly show that when ap-

plied properly GenAI tools can signiﬁcantly improve

productivity even in high complexity cases we also

have to point out some possible limitations. In their

detailed review Tayyba Rasool et. al. has shown that

technological overload may have a negative effect on

productivity (Rasool et al., 2022). This fact may also

be the reason of the quantitatively slightly different

results of the previously mentioned sources.

3 RESEARCH QUESTIONS

Based on the above ﬁndings we posed 4 hypotheses

among 3 main areas related to productivity gain im-

plied by the use of GenAI.

COMPLEXIS 2024 - 9th International Conference on Complexity, Future Information Systems and Risk

130

3.1 The Relevance of Generative AI in

Productivity

While the exact level of the gain implied by the ap-

plication of GenAI proven to be different in different

sources, is seems that a consensus is forming declar-

ing that the use of GenAI has a positive impact on

productivity. Thus we formed our ﬁrst hypothesis as

below:

Hypothesis 1. The application of GenAI tools can in-

crease the productivity in certain areas.

According to the actual state of our studies we have a

limited number of sources to underly this ﬁnding but

even now there are some industrial ﬁelds where ex-

act tools can be identiﬁed for several tasks. A good

example of this is again software development. Ta-

ble 3 in the appendix section presents a collection of

usable GenAI tools for different phases of the Agile

SDLC (See (Sommerville, 2010) - Chapter 3). In the

table classical and Agile SDLC phases are listed non-

necessarily in their time order. We have collected the

related deliverables and successfully identiﬁed the ap-

plicable GenAI tools.

3.2 Measuring Productivity in AI Aided

Software Development and

Education

For different ﬁelds like Software Development, In-

dustrial Management or University Education there

are well-known metrics of productivity with which

it is possible to show the effect of the use of some

new technologies or tools. An interesting questions is

however if we can deﬁne a uniform way of measuring

productivity in these related but in the mean time dif-

ferent areas. According to our review we would like

to underline the following two related hypotheses:

Hypothesis 2. There might exist a uniform metric of

the objective productivity gain.

Hypothesis 3. There should be a methodology which

can be used for measuring the productivity gain.

Our aim in this work is to ﬁnd the uniform metric re-

ferred in Hypothesis 2 and also describe the way how

this metric can be measured in order to provide a pos-

sibility for researchers of this ﬁeld to make a com-

parison to their ﬁndings. And this implies our last

hypothesis related to the uniform property of produc-

tivity.

3.3 Comparison of Productivity Gains

in Industry and Education

Closely related to our previous topic we deﬁned in a

separate hypothesis that according to our expectations

productivity itself measured in different ﬁelds (indus-

try and education) is a comparable measure, so we can

decide where the use of GenAI has more effect.

Hypothesis 4. Productivity/efﬁciency is a universal

measure, that can be applied in both ﬁelds (Indus-

try/Education).

To underline this hypothesis we need to compare the

related ﬁndings of ours and others from the two areas.

As an outcome we would also like to declare in which

area the gain may be bigger.

In order to reach our goals we base our work on

two methods, survey research and analysis of produc-

tivity metrics. While the ﬁrst method can underly our

ﬁrst hypothesis, the others can be revealed by the use

the latter one.

4 METHODOLOGY

In order to have ﬁrst-hand data about the effect of

the use of GenAI tools we base our research on two

sources. i.) At ﬁrst we sent out surveys to three dif-

ferent groups of users of GenAI tools to get a picture

about their subjective opinion and impression about

these. ii.) At second we measured well known pro-

ductivity metrics of the IT industry to have objective

data about the effect.

4.1 A Survey About the Use of GenAI

Tools

Our surveys are being sent out to three different

groups of users of GenAI tools. Namely i.) to uni-

versity Students, ii.) to Academic Lecturers and Pro-

fessors with Executive roles and iii.) Managerial-level

Innovation consultants in Industry.

As a short summary of the structure of all the sur-

veys, in them we deﬁne 5 main topics that we are in-

terested in from three different aspects. By the use of

the results we get from these we expect to be able to

answer questions from Students, Academic lecturers

and IT Industrial Consultants related to

• Section 1: the overall opinion on the productivity

implied by the use of GenAI tools

• Section 2: the aspects of Adoption and Integration

of GenAI

Generative AI for Productivity in Industry and Education

131

• Section 3: the key challenges and opportunities

during the adoption and integration

• Section 4: the possible (and likely uniform) ways

of measuring productivity

• Section 5: the future of GenAI areas.

Technically we have created a ”master survey” that

has been altered in different ways to ﬁt the special

aspects of the different target audiences. As a re-

sult, our ﬁndings coming from different target groups

will become comparable and commonly processable.

In relation with our hypotheses, we suppose that the

GenAI related productivity gain (Section 1) depends

on the other four factors (i.e. the level of integration,

the faced challenges, the way how we measure and

the related expectations).

4.1.1 Identifying SDLC Phase Related GenAI

Enabled Key Use Cases

While results of the surveys may give us an over-

all picture of the use of GenAI in the above men-

tioned ﬁelds, the scope of this survey would be also to

identify the GenAI enabled key use cases within the

SDLC phases including the Requirements Gathering,

Design, Development, Testing, Deployment and Sup-

port such as the key uses cases listed in the Table 1.

4.2 IT Industrial Productivity Metrics

Table 2 presents those agile productivity metrics that

can be used within agile software development lifecy-

cle. We have categorized the metrics into team and in-

dividual level ones and also estimated their expected

importance during the quantitative analysis.

Based on business inside preliminary ﬁndings we

have identiﬁed agile productivity metrics by SDLC

phases and job roles that may potentially be good

measurements for our quantitative analysis. On Table

4 of the appendix section one can note that for each

role we have identiﬁed at least 2 metrics, while for

pure software development related roles several mea-

surable indicators are available.

5 DISCUSSION

In this work we focus on the investigation of produc-

tivity gain implied by the use of GenAI from the as-

pects of industry and education. At the current stage

we have performed a review of the state-of-the-art

ﬁndings to get a picture of the expected effect of it

and also the possible limitations.

Based on them we stated 4 hypotheses, declaring

that according to our expectations i.) GenAI tools can

Table 1: GenAI enabled SDLC Key Use Cases - One aim

of our study is to ﬁnd those use cases that show the most

possible productivity gain when applying GenAI.

Areas Key Use Cases

Key Product

Management

Use Cases

- Ideation and Intake

- Deﬁne and Design

- Develop (linking to SDLC)

- Launch and GoToMarket Planning

- Measure

- Retire

Key Business

Analysis

Use Cases

- Domain and Competitor Analysis

- Requirements Elicitation and Anal-

ysis

- Backlog Creation: Epic, User Story

and Acceptance Criteria Writing

- Process and Data Modelling

- Communications and Stakeholder

Management

- Knowledge Management and Train-

ing

Key Enginee-

ring Use Cases

- API/3rd Party Integration

- Business/Application Logic Imple-

mentation

- Unit Test Coverage

- Code Refactoring

- Code Explanation and Documenta-

tion

- Programming Language Conver-

sion

Key Testing

Use Cases

- Test Case Design and Development

- Test Code Generation and Mainte-

nance

- Test Planning, Execution and Re-

sults Analysis

- Test Case Maintenance and Man-

agement

- Test Data Generation and Manage-

ment

- Test Result Analysis and Defect

Management

Table 2: Proposed agile metrics for measuring the GenAI

productivity gain during software development on team’s

and on individulas’ level.

Priority Metric

Team metrics

High Velocity (Avg. velocity by sprints)

Medium Cycle time

High Lines of Code by Developers (Avg)

High Changed Lines of Codes

High Rework Time

Medium Average Code Review Time

Medium Code Review Failure Rate

Individual metrics

Medium Time in Requirements

Medium Requirement quality

Medium Time in Grooming

High Average Code Review Time

Medium Test Cases Creation

Medium Defect Rate

COMPLEXIS 2024 - 9th International Conference on Complexity, Future Information Systems and Risk

132

boost productivity and ii-iii.) the productivity gain

can be measured by a uniform metric or set of met-

rics. We also state that iv.) productivity itself is a

comparable universal metric that can be used across

these less closely related ﬁelds.

To reason our hypotheses we will perform a sur-

vey based study and quantitative analysis based on in-

dustrial productivity related metrics.

This position paper on the one hand presented the

structure, scope and target of the survey to be used.

On the other hand we have successfully identiﬁed

the investigatable industrial roles, use cases and tools

with their related metrics.

As a next step of our work we will send out the

surveys for our industrial and educational partners.

Meanwhile at our partner comparable projects are

started with and without the use of GenAI tools. Ac-

cording to our expectations we get enough data in the

next 3 − 4 months to be able to start the analytical in-

vestigation.

ACKNOWLEDGEMENTS

The authors acknowledge EPAM for making the

project metrics data available which was essential to

show results based on analysis.

Gergely Kocsis was supported by the project TKP

2021 NKTA of the University of Debrecen Project

no TKP 2021 NKTA 34 has been implemented with

the support provided from the National Research, De-

velopment and Innovation Fund of Hungary, ﬁnanced

under the TKP 2021 NKTA funding scheme.

REFERENCES

Burkitt, F., Zhukov, A., and Razorionov, D. (2023). A call

to action for generative ai. Technical report, EPAM.

Cazzaniga, M., Jaumotte, F., Li, L., Melina, G., Panton,

A. J., Pizzinelli, C., Rockall, E. J., and Tavares, M. M.

(2024). Gen-ai: Artiﬁcial intelligence and the future

of work. IMF Staff Discussion Notes, pages 1–42.

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

framework for university teaching and learning. Inter-

national Journal of Educational Technology in Higher

Education, 20(1):38.

Chan, C. K. Y. and Hu, W. (2023). Students’ voices on

generative ai: perceptions, beneﬁts, and challenges

in higher education. International Journal of Educa-

tional Technology in Higher Education, 20(1):43.

Dohmke, T., Iansiti, M., and Richards, G. (2023). Sea

change in software development: Economic and pro-

ductivity analysis of the ai-powered developer lifecy-

cle.

Eloundou, T., Manning, S., Mishkin, P., and Rock, D.

(2023). GPTs are GPTs: An Early Look at the Labor

Market Impact Potential of Large Language Models.

Papers 2303.10130, arXiv.org.

Harding, W. and Kloster, M. (2024). Coding on copilot -

2023 data shows downward pressure on code quality.

Technical report, GitClear - visited at 15.02.2024.

Hatzius, J. et al. (2023). The potentially large ef-

fects of artiﬁcial intelligence on economic growth

(briggs/kodnani). Goldman Sachs.

Kumar, J. A. (2021). Educational chatbots for project-based

learning: investigating learning outcomes for a team-

based design course. International Journal of Educa-

tional Technology in Higher Education, 18(1):65.

Mishra, A. and Alzoubi, Y. I. (2023). Structured soft-

ware development versus agile software develop-

ment: a comparative analysis. International Journal

of System Assurance Engineering and Management,

14(4):1504–1522.

Musiol, M. (2023). Generative AI: Navigating the Course

to the Artiﬁcial General Intelligence Future. Wiley,

Hoboken, New Jersey.

Rasool, T., Warraich, N. F., and Sajid, M. (2022). Ex-

amining the impact of technology overload at the

workplace: A systematic review. SAGE Open,

12(3):21582440221114320.

Sajja, R., Sermet, Y., Cwiertny, D., and Demir, I. (2023).

Platform-independent and curriculum-oriented intel-

ligent assistant for higher education. International

Journal of Educational Technology in Higher Educa-

tion, 20(1):42.

Sommerville, I. (2010). Software Engineering. Addison-

Wesley, Harlow, England, 9 edition.

Wang, F., King, R. B., Chai, C. S., and Zhou, Y.

(2023). University students’ intentions to learn ar-

tiﬁcial intelligence: the roles of supportive environ-

ments and expectancy–value beliefs. International

Journal of Educational Technology in Higher Educa-

tion, 20(1):51.

Generative AI for Productivity in Industry and Education

133

APPENDIX

Table 3: GenAI enabled activities, deliverables and tools related to the phases of the Agile SDLC. In the table classical and

Agile SDLC phases are listed non-necessarily in their time order. We have collected the related deliverables and successfully

identiﬁed the appliable GenAI tools.

I. Reqs. Gathering II. UX Design III. Architecture Design IV.Delivery Plan

GenAI

accelerated

activity

Personas & Journeys

Stakeholders Interviews

Epics and User Stories

Domain

& General Research

Documents Analysis

Personas & Journeys

Wireframes

Prototyping

Deﬁne design guidelines

& code structure

Deﬁne non-

functional requirements

Deﬁne Quality

Attributes

Generate Delivery Plan

GenAI

assisted

deliverable

Epics & User Stories

High Fidelity Mockup

Application Prototype

Design and

Coding Guidelines

Solution Architecture

Document

Delivery Plan

(tasks, grouping,

sequencing)

Project Risks

& Mitigation Plan

GenAI tool

Azure Open AI GPT

FaceBook LLAMA-2

Amazon Titan

Anthropic’s Claude 2.1

Midjourney

Stable Fusion

DALL-E 2

Hostinger

Durable

Visily AI

WIX

Uizard

Azure Open AI GPT

FaceBook LLAMA-2

Amazon Bedrock

AmazonQ

Amazon Titan

Anthropic

s Claude 2.1

Azure Open AI GPT

Amazon Titan

Anthropic’s Claude 2.1

V. Development VI. Testing VII. Deployment

VIII. Support

& Maintenance

GenAI

accelerated

activity

Translate User Stories

into Gherkin Scenarios

Generate Unit Tests

Coding with Copilot

Code Review

Translate User Stories

into Test Requirements

Generate Test

Speciﬁcation

Generate Functional

tests

Generate Automated

Tests

Test Results Analysis

Update Build

& deployment Scripts

Proactive Monitoring

& Recovery

Bugs Troubleshoo

Proactive Tests Coverage

Refactoring

GenAI

assisted

deliverable

Unit Tests

Source Code

Integration

& Regression Test

Documentation

Test Plan

Test Speciﬁcation

Manual and autoamted

E2E Test

Documentation

CI Scripts

Deployment Scripts

Bug Fixes

Increased Tests Coverage

Refactored Code

Enhanced Maintainability

GenAI tool

Azure Open AI GPT

Google Codey

GitHub Copilot

Amazon CodeWhisperer

Copilot X (Copilot Chat)

Copilot Enterprise

Copilot Workspace

Anthropic’s Claude 2.1

Amazon Titan

Azure Open AI GPT

Google Codey

GitHub Copilot

Amazon CodeWhisperer

Duet AI

Phind

Codeium (Plugin)

Anthropic

s Claude 2.1

Amazon Titan

Azure Open AI GPT

GitHub Copilot

StarCoder

Amazon CodeWhisperer

Duet AI

Phind

Codeium (Plugin)

Anthropic

s Claude 2.1

Amazon Titan

ChatGPT

GitHub Copilot

StarCoder

Amazon CodeWhisperer

Duet AI

Phind

Codeium (Plugin)

Anthropic

s Claude 2.1

Amazon Titan

COMPLEXIS 2024 - 9th International Conference on Complexity, Future Information Systems and Risk

134

Table 4: Identiﬁed Agile productivity metrics by SDLC phases and Job roles.

Agile productivity metric

Velocity (Average velocity by sprints

Cycle time

Time in Requirements

Requirement quality

Time in grooming

Lines of Code by Developers (Avg)

Changed Lines of Codes

Rework Time

Average Code Review Time

Code Review Failure Rate

Test Cases Creation

Defect Rate

SDLC Phase Job role

Requirements Gathering Business analyst X X X X

UX Design UX Designer X X

Architecture Design Software Architect X X

Delivery Plan

Project manager, Delivery

manager, Scrum master

X X X

Development Developer X X X X X X X

Testing

Test automation engineer,

Manual test engineer

X X X

Deployment DevOps engineer X X X X X X X X

Support DevOps engineer X X

Generative AI for Productivity in Industry and Education

135