INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026

Comparative Analysis of Web Design Performance: AI‑Assisted

Tools Versus Traditional Figma‑Based Design

ABSTRACT

This study compared the performance of traditional Figma workflows and AI-assisted Figma workflows in

developing high-fidelity UI/UX prototypes among BSIT students of Quezon City University during the

Academic Year 2025–2026. Using a descriptive-comparative within-subjects design, the study evaluated both

workflows in terms of Design Quality (Functional Suitability, Usability, and Performance Efficiency) and

Creative Output (Creative Autonomy, Innovation, and Ease of Workflow). Data were collected from 60

purposively selected respondents through a researcher-developed questionnaire and analyzed using weighted

mean, standard deviation, Wilcoxon Signed-Rank Test, and Spearman’s Rank-Order Correlation. Results

showed that both workflows received positive evaluations; however, the traditional Figma workflow consistently

obtained higher ratings. Significant differences were found in Functional Suitability, Usability, and Creative

Keywords: creative output, design quality, prototyping performance, usability, workflow efficiency, web

designs, artificial intelligence, Figma

INTRODUCTION

In the era of rapid digital transformation, web design has evolved from a purely manual process into a

collaborative interaction between human creativity and computational intelligence. As 2026 continues to witness

advancements in Artificial Intelligence (AI), its integration into graphic and web design workflows has become

central to modern creative practices, significantly transforming how digital interfaces are conceptualized and

developed (Chamorro-Premuzic, 2023). Industry-standard platforms such as Figma have expanded beyond

traditional vector-based editing and prototyping, incorporating AI-assisted and automated functionalities that

streamline the design process. According to Staiano (2020), proficiency in these advanced design platforms is

efficiency by automating repetitive tasks and generating layouts through generative models (Sun, 2026).

Despite these technological advancements, concerns continue to arise regarding the limitations of AI and its

implications for human-centered design. Mester and Toth (2025) highlight that although AI contributes

significantly to productivity and automation, it also introduces challenges associated with accountability,

incorrect decision-making, and privacy concerns. As society transitions toward Web 4.0, the relationship

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026

between humans and machines is becoming increasingly interconnected; however, this integration also raises

issues related to algorithmic bias and digital inequity, potentially excluding marginalized communities from

equitable digital experiences (Mester & Toth, 2025).

Moreover, while AI enhances operational efficiency, it often lacks the ability to fully understand complex human

emotions, behaviors, and contextual nuances. This creates a significant challenge in modern web design, where

research gap persists regarding the direct comparison between AI-assisted workflows and traditional Figma-

based design approaches. Existing literature primarily focuses on the technical capabilities of generative tools

and theoretical discussions in design education, with limited emphasis on practical usability assessments.

Quiñonez-Ku et al. (2024) stress the importance of evaluating prompt-driven designs against established smart

mobile application design standards. It remains uncertain whether the efficiency offered by AI-driven automation

compromises design quality, usability, and the intentional creativity of designers, which is often obscured by the

technical “black box” nature of AI systems.

This study seeks to examine the comparative effectiveness of human-led creativity and AI-assisted automation

in web design performance. Specifically, it aims to determine whether generative AI tools significantly enhance

creativity and productivity or whether they introduce structural limitations that hinder the development of high-

prototypes; however, these outputs often require extensive human refinement to satisfy professional usability

intentionality and creative control. Similarly, Mester and Toth (2025) argue that AI-assisted systems are most

effective in automating repetitive layout and structural tasks, while the core elements of creativity, originality,

and emotional connection remain fundamentally dependent on human designers.

Nevertheless, AI-generated designs continue to face structural and contextual limitations. Saita (2024) explains

that AI frequently struggles to interpret cultural nuances and complex human-centered considerations, leading

to interfaces that prioritize appearance over practical usability. Therefore, expertise in traditional design

methodologies remains essential. Staiano (2021) further emphasizes that deep knowledge of interface

prototyping tools such as Figma, particularly its advanced manual controls, is necessary to transform basic AI-

The significance of this study lies in its potential contribution to both academic and professional design practices.

By identifying the strengths and limitations of human-centered and AI-augmented design methodologies, the

research aims to provide a foundation for developing a hybrid design framework that maximizes efficiency while

preserving the creativity, originality, and human insight essential for effective digital communication.

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026

Statement of the Problem

This study aims to evaluate and compare the effectiveness of AI-assisted prototyping tools versus traditional

Figma-based workflows in terms of design performance, prototyping quality and user centered usability.

Specifically, it seeks to determine if the automation provided by generative AI enhances creative efficiency or

introduces structural limitations that affect the final high-fidelity output.

Specifically, this study seeks to answer the following questions:

assisted and traditional workflows in terms of the aforementioned variables?

What is the level of creative output and professional satisfaction among designers using these methods in terms

of:

Is there a significant relationship between the level of design quality of high-fidelity prototypes produced by

both AI assisted and traditional Figma workflows and level of creative output?

Related Studies

The continuous advancement of Artificial Intelligence (AI) and Model-Driven Development (MDD) has

consistently acknowledge the substantial efficiency benefits offered by AI and MDD, debates remain regarding

their effects on usability, structural quality, creativity, and the preservation of essential human design skills. A

synthesis of current literature suggests that while AI significantly accelerates prototyping and development

processes, comprehensive comparative evaluations remain necessary to determine whether AI-assisted

workflows can truly achieve the same level of usability, structural integrity, and creative autonomy associated

with traditional human-centered design approaches.

Traditional UI/UX design processes require extensive iterative development, detailed planning, and

multidisciplinary technical expertise (Malik et al., 2023). These requirements often result in prolonged

development cycles and increased susceptibility to repetitive design errors. To address these challenges,

researchers and practitioners have increasingly adopted automated and AI-driven frameworks. Studies on

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026

developers utilizing automated infrastructures can achieve productivity levels approximately 5.8 to 11.7 times

greater than those relying on conventional manual development approaches (Priefer et al., 2021). Similarly,

generative AI technologies, particularly Large User Interface Models (LUIMs), demonstrate significant potential

and-own” UI development practices (Priefer et al., 2021).

showing a reduction in code style violations by approximately 2.4 times through automated systems (Priefer et

al., 2021). On the other hand, recent evaluations of AI-generated user interfaces reveal notable shortcomings in

addressing essential usability dimensions such as effectiveness, efficiency, learnability, and user satisfaction

(Namoun et al., 2024). Existing AI design tools often lack sufficient understanding of established UI/UX

principles, user-centered methodologies, and contextual human empathy (Namoun et al., 2024). This contrast

demonstrates that while automation can strengthen structural consistency and technical standardization, AI-

generated outputs still depend heavily on human intervention and intuition to achieve meaningful usability and

user-centered functionality (Priefer et al., 2021; Namoun et al., 2024).

The integration of Generative AI (GenAI) into design workflows also introduces significant psychological and

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026

substantial human guidance (Namoun et al., 2024). Additionally, concerns regarding creativity, design intuition,

skill preservation, and professional autonomy continue to shape discussions surrounding AI adoption in design

practices. Although prior studies have validated the effectiveness of comparative experimental methods and self-

need for a controlled and comprehensive comparative study focused specifically on evaluating prototype quality,

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026

The target population of this study consisted of BSIT students enrolled at Quezon City University during the

Academic Year 2025–2026. Based on institutional estimates, the BSIT population comprised approximately

2,882 students. BSIT students were selected as respondents because their academic training and coursework

concerning creative autonomy, innovation, and divergent thinking within AI-assisted design environments.

Prior to data collection, the questionnaire underwent face validation by faculty experts to assess the

appropriateness of the language, structure, and contextual relevance of the instrument for third-year BSIT

grounding in established UI/UX evaluation frameworks and existing literature on AI-assisted design assessment

(Namoun et al., 2024; Lively & Hutson, 2024).

Statistical Treatment of Data

The data collected from the 60 respondents were analyzed using statistical techniques appropriate for ordinal

data derived from a four-point Likert scale. The selected statistical methods were aligned with the research

questions and objectives of the study.

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026

Weighted Mean and Standard Deviation were utilized to determine the levels of Design Quality and Creative

Output for both the traditional and AI-assisted workflows across all evaluation criteria. These statistical measures

addressed Research Questions 1 and 3 by describing the central tendency and variability of participant responses

Likert-scale mean analysis (Lively & Hutson, 2024).

paired ordinal data in situations where assumptions of normality cannot be guaranteed (Field, 2018). The

selection of this statistical procedure is also supported by the comparative framework utilized by Fu et al. (2024)

in evaluating AI-assisted and non-AI design environments.

To examine the relationship between Design Quality and Creative Output across both workflows, the

Spearman’s Rank-Order Correlation Coefficient (ρ) was used in response to Research Question 4 (Field, 2018).

Spearman’s correlation is appropriate for ordinal datasets because it does not require interval-level measurement

or normally distributed variables (Field, 2018). The interpretation of correlation strength followed Cohen’s

correlation (ρ ≥ 0.50). The use of this relationship analysis is theoretically supported by the findings presented

1.76–2.50 (Low), and 1.00–1.75 (Very Low).

Table 1. Level of Design Quality: Traditional Figma vs. AI-Assisted Workflow (n = 60)

organize interface functions, properly align design components with project requirements, and effectively

evaluate the functional appropriateness of each element within the interface.

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026

In comparison, the AI-assisted workflow obtained a mean score of 3.01 (SD = 0.62), interpreted as High.

Although respondents generally agreed that AI-assisted tools were capable of generating functionally appropriate

outputs, the level of confidence was noticeably lower than that observed in the traditional workflow. The higher

intended functional requirements of the project.

Usability

With respect to Usability, the traditional Figma workflow achieved a mean score of 3.41 (SD = 0.46), interpreted

as Very High, whereas the AI-assisted workflow obtained a mean score of 3.12 (SD = 0.62), interpreted as High.

The higher rating for the traditional workflow indicates that participants perceived manually designed interfaces

as more effective in supporting human-centered design principles, maintaining visual hierarchy, and ensuring

intuitive interaction design. Participants also believed that traditional workflows provided greater flexibility in

applying established UI/UX guidelines and incorporating user empathy into the design process, factors identified

by Namoun et al. (2024) as essential components of effective interface usability.

assisted design systems still face limitations in addressing critical usability dimensions such as effectiveness,

learnability, and user satisfaction. The results imply that while AI-assisted tools are capable of producing usable

interface prototypes, human involvement remains essential in refining usability and ensuring user-centered

interaction design.

Performance Efficiency

In terms of Performance Efficiency, both workflows produced relatively similar results. The traditional Figma

workflow obtained a mean score of 3.28 (SD = 0.46), interpreted as Very High, while the AI-assisted workflow

achieved a mean score of 3.20 (SD = 0.51), interpreted as High. The minimal difference between the two scores

suggests that participants perceived both workflows as effective in supporting efficient interface development.

The strong performance of the AI-assisted workflow may be attributed to the automation capabilities of

workflows among all measured Design Quality dimensions. This finding suggests that AI-assisted tools have

become increasingly effective in improving workflow speed and reducing repetitive design tasks. The results

further support the findings of Lively and Hutson (2024), who concluded that AI-assisted technologies

significantly enhance workflow efficiency and accelerate design production processes.

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026

Significant Difference Between AI-Assisted and Traditional Workflows

Research Question 2 examined whether statistically significant differences exist in how respondents assessed

both workflows across all six criteria. Given the ordinal nature of Likert-scale data, the Wilcoxon Signed-Rank

Test was used at a significance level of α = 0.05 (Field, 2018).

Table 2. Result of the Wilcoxon Signed-Rank Test Conducted

Table 2 presents the results of the Wilcoxon Signed-Rank Test conducted to determine whether significant

differences existed between the traditional Figma workflow and the AI-assisted workflow across the six

evaluation criteria. Significant differences were identified in Functional Suitability (p = 0.0002), Usability (p =

0.0049), and Creative Autonomy (p = 0.0175), where the traditional workflow consistently obtained higher mean

scores than the AI-assisted workflow. These findings suggest that participants perceived traditional manual

design processes as more effective in accurately translating functional requirements, maintaining usability

standards, and preserving creative control. The results support the findings of Namoun et al. (2024), who

emphasized that AI-generated interfaces still struggle to independently satisfy complex usability and functional

requirements without substantial human intervention. Similarly, the lower rating for AI-assisted workflows in

Creative Autonomy aligns with the observations of Fu et al. (2024) and Akça (2026), who noted that AI-assisted

repetitive design tasks. However, the results also suggest that traditional workflows remain sufficiently efficient,

particularly for users already familiar with manual Figma design processes. Similarly, the lack of significant

difference in Innovation implies that both workflows are capable of supporting creative idea generation and

design exploration.

Overall, the findings demonstrate that while AI-assisted workflows provide competitive advantages in efficiency

and workflow convenience, traditional Figma workflows continue to perform better in areas requiring human-

centered judgment, usability refinement, and creative autonomy. The results suggest that AI tools currently

function more effectively as supportive technologies rather than complete replacements for human designers.

Consequently, the integration of AI into UI/UX design may be most beneficial when combined with human

creativity and manual oversight, allowing designers to leverage automation while maintaining control over