VIETNAM NATIONAL UNIVERSITY, HANOI INTERNATIONAL SCHOOL STUDENT RESEARCH REPORT MEASURING THE ACCEPTANCE OF USING BING AI TECHNOLOGY IN THE PACKAGING DESIGN OF AI USERS AND CONSUMERS
INTRODUCTION
Research Motivation and Background
The strong development of the Fourth Industrial Revolution (or Industry 4.0) has marked a significant and unfamiliar turning point in production activities, where technology is rapidly advancing to a level that is difficult to control (Ken Rolfes, 2019) Prominent advanced technologies in the context of Industry 4.0 include the Internet of Things (IoT), Artificial Intelligence (AI), and automation According to a report by Maximize Market Research, by 2023, Industry 4.0 technologies are predicted to have a profound impact on organizations worldwide, with AI being the main driving force of the transformation (Global Industry 4.0 Market, 2024) AI refers to the ability of a machine to perform human-like functions such as learning, reasoning, and problem-solving It enables computer systems to learn from experience and be creative based on initial data (Banitaan et al., 2023) With the advancements in AI, a popular term that is mentioned is Chatbot, which is continuously infiltrating every aspect of human life According to Bhoir, a Chatbot is an AI program that simulates conversation with users through messaging applications, mobile apps, or phones (Bhoir et al., 2022) Chatbots not only help users access information quickly but also optimize strategies and plans, enhance content creativity, and personalize messages (Quach, 2022) Therefore, the application of AI Chatbots, especially in the field of digital marketing, has become an important part of the business strategies of many companies Chatbots are seen as a completely new technology, as they have been upgraded to the point where they can replace faces, mimic voices, brand products, and are called
"Deep fakes," as Hanna writes, as all of this can be done with an impressive speed by simply pressing a button on a computer (Hanna & Arts, 2023)
In addition, based on the current market economy, building a brand image is essential for businesses to survive in harsh economic conditions, to have long-term sustainability, and to maximize their strength and positioning among other brands (Goad,
1999) Developing a brand image is particularly important in Asia in general and Vietnam
8 in particular, as companies have to face the challenge of the presence of famous international brands from abroad and the perception of the Vietnamese people that Vietnamese brands are inferior (Bernd H Schmitt, 1994; Jacob, 1993; Schutte, 1998) To create this, according to Hutton, stimulating visual perception is an important part of any brand-building strategy It refers to factors such as logos, signage, packaging, product design, advertising, and websites (Hutton, 1997) Indeed, Krishna's research (2012) confirmed that marketing through visual stimulation using product packaging will convey the meaning of the brand as well as the abstract attributes of the product (Krishna, 2012)
It will affect expectations of the product experience - its appearance, sound, feel, smell, or taste Previous studies have shown a high correlation between product design quality and the financial performance of companies (Hertenstein, 2001) This further demonstrates the important and necessary role of the Design department in building a brand for the entire business The idea and design representing a business will be carried out by a team of experts who perform complex processes to create innovative and unique designs It can be said that the process of generating ideas and how to create an image that closely aligns with the initial idea is a long-term and resource-intensive process
Therefore, thanks to the strong development of Chatbot Open AI technology to minimize the time in the design process, users can use terms (called "prompts") to rephrase ideas for AI to learn and generate creative designs based on the provided data (Sami, 2022) The AI technologies that can do this include Bing AI, Midjourney, Dall-E, GPT-3 (Chat GPT), According to Hanna's research (2022), after surveying measuring the level of recognition and acceptance of AI in enriching advertising activities as well as an effective and useful solution, she concluded that using Midjourney produced impressive results and is a design solution for all fields, saving time and effort for designers (Hanna & Arts, 2023)
Packaging design significantly influences customer brand perception and attitudes Numerous studies have substantiated this correlation, highlighting the crucial role of AI and automated packaging design technologies in enhancing customer experiences.
9 are still relatively new and have not been extensively evaluated to determine whether applying these AI technologies actually saves users time, effort, and truly enhances customer brand attitude This is still a mystery Therefore, our research team has decided to conduct this study to measure the acceptance of using Bing AI technology in product packaging design by users and customers.
Research Objective, Questions and Scope
By measuring the effectiveness of Bing AI technology in automated product packaging design, this study aims to forecast user acceptance and trust in its implementation Theoretically, it will determine the proportion of users and customers expressing interest in Bing AI for packaging design, assessing its acceptance level within the business environment Practically, the research will outline real-world applications of Bing AI, providing businesses with insights for optimizing their strategies based on measurable data.
To achieve this objective, the study will address the following research questions:
- Do users genuinely accept automation in AI-driven product packaging design?
- What are the most important factors in demonstrating a positive relationship between users and the features of Bing AI technology?
- Are consumers willing to purchase products with designs created by AI?
- Does AI-driven design contribute to efficiency in production or is it merely a reference idea?
- What initiatives can be implemented in Vietnam to promote user acceptance of AI- generated automated designs?
However, due to limitations in the scope of the research, particularly focusing on residents in Hanoi in general and students at the International School - Hanoi National University specifically, the study may not fully reflect all viewpoints and the actual market situation.
Research Method
After investing time in researching previous literature and understanding the community's needs related to this novel topic, the research team has decided to apply a quantitative method to collect information and data, thereby providing a basis for the research questions This process has been divided into three main stages
To understand user perceptions of AI design technology, we assessed relevant literature and Hanna's (2022) survey questionnaire This informed our development of a standardized questionnaire in both English and Vietnamese, structured into four main sections and three sub-sections The questionnaire, comprising over 30 questions, gathered participant information and explored their opinions on AI design technology and related factors The detailed questionnaire is available in Appendix A of the research report.
Next is the second stage - the data collection stage We collected data from approximately 249 participants, mainly from Hanoi and a few from Ho Chi Minh City and other provinces The data collection process took place from early March to early April
2024 We distributed survey links to over 300 students at the International School - National University of Hanoi, as well as reaching out to residents of Hanoi and a few individuals working in the design field
In the third stage, after synthesizing the data, we proceeded to clean the data by removing invalid, irrelevant, or redundant responses, reducing the total number of participants to 235 valid final responses to ensure quality After the data was cleaned, we
11 performed data analysis and descriptive statistics to summarize and gain a deeper understanding of the distribution of variables in our dataset We further examined and merged variables correlated with factors in the research model, then conducted in-depth analysis based on indices regarding time, AI design, perceived attractiveness, and brand attitude These indices revealed the level of acceptance of using AI technology in product packaging design
After the analysis of this questionnaire, we will evaluate the model using SMARTPLS software SmartPLS is known as a leading software tool in applying and approaching the PLS (Partial Least Square) method in linear structural estimation models Specifically, the software has the function of estimating SEM (Structural Equation Modeling) - estimating the parameters of the model based on the variance-based matrix According to Hair & colleagues (2016), PLS-SEM applies ordinary least squares (OLS) regression techniques with the goal of minimizing errors (i.e., residual variances) of dependent variables (Hair, 2016) We chose SmartPLS for its lightweight and user-friendly nature, while still ensuring effectiveness in model estimation By using this software, we do not require a large sample size like other tools, which is suitable for the scope of our study We used SmartPLS to measure the factors influencing the acceptance of AI technology in product design, including time, AI design, attractiveness, brand attitude, demographics, and other factors The results are presented through tables and charts, from which we draw conclusions based on statistical analysis of the data.
Structure
This research will consist of six chapters The first chapter provides an overview of the context, research objectives, scope, methodology, and structure of the study, including the objectives, research questions, and techniques used to carry them out
In Chapter 2, the paper will present a set of concepts used to explain and describe the context of automated AI design technology and delve into Bing AI technology, the
12 steps involved in using this technology to assist users in designing product packaging These concepts are part of the Theoretical Framework content
Moving on to Chapter 3, literature on time, user AI design using Bing AI, perceptions of attractiveness, as well as attitudes towards the brand of customers when viewing advertising products designed by Bing AI will be examined in depth We examine the concepts and components of the effectiveness that Bing AI brings to users and customers This chapter also highlights the evaluations of previous studies on the relationship between the benefits of AI for users and customer brand perceptions towards the acceptance of Bing AI technology The development of hypotheses and research gaps will be addressed at the end of the chapter
The research technique is fully described in Chapter 4 along with the steps related to data collection, data cleaning, exploratory data analysis, and square regression method analysis
In Chapter 5, the results of data analysis are presented and discussed based on the research questions and hypotheses
Finally, in Chapter 6, proposals based on evidence drawn from the research results will be made to clearly determine the extent to which people will accept the use of Bing
AI automated design technology in their lives, especially in the business world Practical and theoretical implications as well as the limitations of the paper will also be included in the chapter content
THEORETICAL FRAMEWORK
The historical of AI Image Creator Models
For centuries, humans have embraced the idea of intelligent machines with magical features that can be applied to life as in myths or through science fiction films In Greek mythology, there is a reference to a giant copper self-propelled machine that protects the island of Crete Or, like in the science fiction movie that mentions the HAL computer that can control an entire spacecraft operating in outer space From the ideas and the desire to realize those ideas, artificial intelligence (AI) technology was born, from skepticism about its surrealism AI has now become part of our daily lives, not only helping to answer basic questions, but it also helps us increase work performance, predicting possibilities that may occur in the future And most recently, in the form of genetic AI, this tool also helps us realize text into images, making art creation much easier to reach any audience
Artificial intelligence (AI), to specify its concepts, can be understood in literal terms
- the way machines operate simulated by human intelligence (AI tạo sinh so với các loại
AI khác., 2023) A typical example of how AI works, also known as machine learning, is how AI plays automatic games that simulate the way humans play chess - a project by the University of Manchester first announced in 1951 Or more commonly, for example, Siri and Google Assistant are pre-installed on your smartphone Just call these tools and command them, they can perform a number of tasks directly and are compatible with many different languages This is called a form of Narrow AI operation According to Sergey.N
(2023), initially models like the naive Bayes classifier were used to create text, but they were not suitable for image creation due to the complexity of image data Automatic regression models such as Pixel-CNN and Pixel-RNN have been developed to generate each pixel based on previous pixels, but they are not effective for generating high- resolution images due to the inability to parallel the process
In recent decades, AI has become increasingly powerful in image recognition and creation, with models such as deep neural networks leading in the creation of high-quality images and even artworks According to the synthesis of (Notomoro, 2023), the first and most popular original models of AI imaging include: Generative Adversarial Networks (GANs), this model uses two opposing neural networks to create new images while Variational Autoencoders (VAEs) focused on encoding and decoding the data to create new images with the same structure as the training data, and finally is Autoregressive Models: These models predict each pixel of the next image based on the previous pixel In addition, companies with the potential to develop AI such as Microsoft or OpenAI and so on have developed their own image-generating AI models, often based on or expanded from these original models to create more powerful and easy-to-use image generating tools
In fact, these models are constantly being improved and updated with new features to generate accurate, high-quality images that meet a wide range of user needs
The working mechanism of an image-generating model from a text description requires a multimedia model to transform text and images into the same space AI converts descriptive text into images through a process called deep learning, using advanced machine learning models such as deep neural networks The process of creating an image takes place in four steps: Model training: AI models are trained on large datasets containing corresponding pairs of text and images In the course of training, the model learns how to associate words and phrases with specific images; Understanding the context: When receiving the text description, the model uses algorithms to analyze and understand the context, meaning and descriptive elements in the text; Image Creation: Using the learned knowledge, the model creates a new image that reflects the elements described in the text This process can include selecting from a latent space or creating each pixel of an image; Optimization: During image creation, the model can be continuously adjusted to improve the quality and accuracy of the image based on the criteria learned (Ramesh et al., 2022) has represented that process of DALL-E through the following model:
Figure 1: Process of Dall-E model
OpenAI's DALL-E was the first model to claim to this goal with good quality, using variational autoencoder and Transformer This model has the ability to create diverse images, including creating personalized versions of animals and objects, logically combining unrelated concepts, displaying text, and applying transformations to existing images Later models such as DALL-E 2, Midjourney and Stable Diffusion have evolved further and surpassed DALL-E in terms of the ability to generate images from text descriptions
Specifically, the DALL-E 2 and DALL-E 3 are the next versions of the DALL-E, with the ability to create, edit, and combine images from text descriptions DALL-E uses a variation of the GPT-3 model with 12 billion parameters to create a diversified image from the text description The DALL-E model is very popular when adopted by ChatGPT Plus and Bing AI Image Creator According to (Horsey, 2023), They also have a difference in image-making style, which is that Bing Image Creator can create more detailed and grinding images, while Dall-E 3 on ChatGPT Plus stands out in producing high-quality images At the same time, there are limitations in content creation, for DALL-E 3 on ChatGPT Plus, limiting the ability to create darker and more angular images, while Bing Image Creator may be more suitable for these tasks Midjourney software is not open
16 source either, and it is known for its ability to create higher artistic images than other software, with the latest version of V6 improving the ability to generate images based on input commands and can contain text in images Finally, Stable Diffusion, the only open- source model available today, allows users to customize and build their own AI tools According to what they presented on Github, this model uses data from LAION-5B and previous research projects such as OpenAI's ADM and denoising-diffusion-pytorch These are key components that enable Stable Diffusion to generate high-quality images from text descriptions, which are much different from other software It allows users to run directly on personal computers, bringing more control and customization, even creating personal
AI tools In addition to trying some photo and video editing tools, publishing designers also apply AI models to their software so that they can automatically identify many features in areas of images that need to be deleted, edited or… Stable Diffusion’s model by (Rombach et al., 2022) shown on the illustration model as follows:
Along with great features, AI Image Generator still has shortcomings in recognizing the meaning of a user's sentence in different languages, resulting in the more images are generated, the more the images are learned after each time and the more they deviate from
Artificial intelligence (AI) technology is rapidly advancing, driven by the provision of vast datasets to improve neural network capabilities However, this progress raises concerns about AI's potential to manipulate images and disseminate misinformation, creating opportunities for targeted online campaigns.
Bing AI Image Creator: Overview
After integrating AI into Bing chat both on the Microsoft Edge browser and directly on the Skype app, Microsoft realized the potential exploitation of artificial intelligence technology has changed the user experience of searching for information on the browser Searching becomes more interesting when it gives the user the feeling of being in a conversation, through a message-like display, information expressed in more detail than the traditional search but in a form of entertainment or creative inspiration for the user According to experts at Munch:studio, human’s brain is very much influenced by visual images, it allows us to communicate easily, and thus receive better information (Mehdi,
2023) has come to the conclusion that images are one of the most searched categories on this platform To further that potential, in March 2023, Microsoft announced that it had officially launched Bing Image Creator, a platform capable of generating images with text- only reminders Users can use this tool directly in a chat with the Skype Bing virtual assistant, just using your own expression to describe the image you want to create Microsoft's goal is to integrate both text content and create images on the same interface to enhance the user experience Now, instead of users being restricted to searching only available images that have been posted online, they are virtually allowed to search and satisfy their creativity with their rich imagination and relevance
According to (Mehdi, 2023) - Microsoft Corporate Vice President & Consumer Chief Marketing Officer, in the process of providing Bing with a preview test for some users, Microsoft has collected and received their actual feedback to find out how users use it correctly to their orientation, thereby improving their experience, along with improving
18 the output image quality And this tool is compatible with both desktops and mobile devices, from directly on the official website at bing.com/create to the Microsoft Edge search engine and also on the Skype chat window This is considered the most outstanding feature of this tool compared to other software when it comes to providing the most convenient user experience, without the need for a third party, users can simultaneously use the search for questions and create images on the same platform
Bing Image Creator uses an advanced version of the DALL-E model from OpenAI to make the basic model for its software, so Bing's image quality will also carry the relative characteristics of this model, the quality of the image makes it feel like a real picture, the image creates more vivid, can understand and logically combine irrelevant words Bing Image also has great potential when users of Microsoft platforms get access to Bing Image Creator for free, which is of no inferior quality to paid software At the same time, the working principle of this model is to collect the visual and text data of the user as input data to perform activities that create new images from linking, linking information, that is, the more information in the system will learn more and give the image more accurately to the user's wishes By giving users free access, the model will be more accessible
2.2.1 How to use Bing Image Creator
To access Bing Image Creator, users require a Microsoft account, which can be created through various Microsoft software or directly on the Bing Image Creator website This account serves as a storage space for generated images, ensuring they are not lost when the browser is closed While the Creator allows free text-to-image creation, using booster moves speeds up the process Regular accounts can continue creating images after boosters are depleted, but with increased wait times, while business or school accounts offer different benefits.
19 times, for school or business accounts after the enhancement is expired, that account will be restricted to creating additional images until additional ones are granted After registering and successfully logging into the website, the system will redirect the user to the main desktop interface
The picture below shows how to start accessing and using Bing Image Creator:
Figure 3: Bing Image Creator interface
Bing Image Creator features a writing prompt bar, fixed images with suggested prompts, and "Creations" for storing up to 20 user-generated images for 90 days To ensure permanent storage, download images after creation The image creation feature syncs with Skype chat, allowing users to access created images on both the website and Skype.
And for the built-in version of Bing Image Creator on Skype, to create a picture, you need to not only have a Microsoft account, but also connect to Bing chat on this app When successfully logged in to Skype, Bing chat will be suggested for the user to try, it will appear at the very first chat on the Skype interface, the user needs to click on to start the chat, wait when Bing has been successfully set up to Skype is already able to use Bing as a search engine that combines images conveniently Images created via Skype can be viewed in the chat history section, it can store more images, and the storage time is longer than the website version
Here's the interface that uses the Bing Image Create feature built into the Skype chat framework:
Figure 4: Bing Image Creator on Skype interface
To use Bing AI Image Creator, users need to type in the description bar of the reminder they want to create The more detailed the reminder (context, subject location, color intensity, etc.) the more accurate it is Once the image has been created, users can improve the image by describing deleting, adding, shifting, etc commands of objects in the image until the image is perfectly suitable Users can add keywords related to drawing
21 such as “create an image” or “draw a picture” along with a description of the idea they want to draw before describing the content of the image as a reminder in the conversation Each time you type a prompt, the system will give out four different images according to the request described, if the user does not specifically describe the drawing style of the image then the four images will be designed in the same style by default, the default size of Bing Image Creator is square and cannot be customized
2.2.2 Bing AI Image Creator Design Experiment
In this section, the research team will present how the research team applies Bing
Bing AI Image Creator's packaging design capabilities were tested on renowned brands like Dove, Heineken, Nike, and Heinz, to showcase its creative versatility To maximize AI's potential, the team incorporated imaginative descriptions that these brands had not previously released These descriptions guided the AI's design process, enabling the exploration of unique and captivating product packaging concepts.
In the first step, the research team requested Bing to create an image with the theme of creating a new product collection for the Dove brand, the product range includes basic items such as shampoo and conditioner, body wash, and exfoliating scrub The research team also seeding prompt detailed product components which are argan oil - a commonly used ingredient in hair care products, and honey, with an illustration on the packaging of a drop of essence in the middle of the bottle, the background is an illustration of those components, specific prompt as follows: “Draw a new Shampoo packaging of Dove brand
22 with argan oil The bottle body is white, with the logo of Dove brand on the top of the bottle, a photo of argan oil essence in center, modern style but close to nature”
After receiving the request from the research group, the Bing Image Creator software automatically transformed them into sentences as follows: "A new shampoo packaging of Dove brand with argan oil The bottle body is white, with the logo of Dove brand in the top of bottle, in the center have a photo of argan oil essence, modern style but close to nature" Below is the result that the research group collected from the prompt
Figure 5: Dove shampoo packaging design by Bing AI Image Creator from 1st prompt
LITERATURE REVIEW
Time
In the context of today's digital age, many businesses have enhanced their brand recognition by providing customer experiences to maximize satisfaction and interaction
To meet this demand, two emerging fields have become the focal point of technological advancements: User Experience (UX) design and User Interface (UI) design These two fields are currently thriving in the "Golden Age" of the Internet According to Browne
The importance of enhancing customer experience through effective UX/UI design has become increasingly evident, positively impacting PR efforts As a result, businesses highly value professionals skilled in these fields This presents a significant challenge for designers to continuously improve their expertise in UX/UI to meet the evolving demands of the industry.
Indeed, for designers, it is not only necessary to understand technology and customer needs but also to possess creativity and innovation in their style to contribute to the success of a business Han (2022) mentioned a term when analyzing the challenges faced by designers, known as "Design Thinking" (Han, 2022) According to her, design thinking in business is an ideology based on the working process of designers to outline design stages Its purpose is to provide all experts with a standardized innovative process to develop creative solutions This is an iterative process in which the design team must be flexible to understand customer needs and generate solutions to address those needs To further understand the application of design thinking, the Hasso Plattner Institute of Design at Stanford University introduced a method called the "Design Thinking Process" (DT) (IxDF., 2016) This process consists of five stages: Empathize, Define, Ideate, Prototype, and Test The stages of DT are illustrated as follows:
Figure 11: The Design Thinking Process
According to researchers at the institute, this process does not always follow a strict sequential order; they can run in parallel, out of order, and repeat when necessary
(1) Emphasize (2) Define (3) Ideate (4) Prototype (5) Testing
During the "Ideate" stage, design teams synthesize customer insights to generate diverse solutions Transitioning to the "Prototype" stage, they transform these ideas into tangible forms Through testing, iteration, and refinement, designers ensure customer satisfaction and translate abstract concepts into feasible products, services, or experiences This pivotal stage plays a crucial role in bridging the gap between conceptualization and practical implementation.
If this stage relies solely on human effort, meaning the design team uses traditional applications like Adobe Illustrator, Photoshop, etc., to translate abstract concepts into visual product representations, how long would it take to complete the design? According to Padmasiri et al., they found a high demand for product packaging design (Padmasiri et al., 2023), and this process requires a significant amount of manpower, time, and resources (Yueyi, 2019)
This can be considered one of the main reasons why businesses are shifting towards using Artificial Intelligence (AI) to reduce design costs and optimize time while ensuring effectiveness in attracting customer attention through conveying brand messages and enhancing product appeal (Alhamdi, 2020); (Azzi, 2012) AI utilizes data-driven methods and algorithms to generate personalized packaging design templates (Liu, 2022; Shi, 2022) The design team only needs to focus on understanding customer needs and, as mentioned in the background section, provide information to AI through data analysis AI will then generate diverse packaging design models according to consumer requirements To illustrate the advantages of AI more clearly, consider the following examples:
- In a study by Chen et al in early 2023, a company in Japan used PLUG AI in product packaging design and realized that they could complete 1,000 packaging design templates in just 1 hour based on materials and images from their design team Or in the United States, the VIZIT platform has the ability to predict market acceptance of packaging designs based on intelligent databases (Chen et al., 2023)
- Previous studies have also indicated a positive relationship between designers and the use of AI technology Specifically, after analyzing the questionnaire responses of UX experts
30 in Sri Lanka using a snowball sampling method, Padmasiri and colleagues found that during the ideation and prototyping phase of the design thinking process, 79% of participants received support from AI tools in customizing web interfaces by using images generated by AI (Padmasiri et al., 2023) In another study by Hanna on the use of AI technology Midjourney in product advertising, she included questions in her survey about the time taken to evaluate the time efficiency of the Midjourney Dall-E technology for designers After data analysis, she concluded that while 85% of participants were aware of this application, only 35% of them actually used it due to its novelty The study's conclusion indicated that designing packaging through AI technology saves time and effort for designers (Hanna & Arts, 2023) Furthermore, Hanafy's (2023) study has shown a positive correlation between the use of Midjourrney, Dall-E2, and StableDi technologies in the technical design process
In addition to the tools mentioned in the above examples, Microsoft recently introduced Bing Image Creator, a concept similar to Midjourney, Dall-E, etc
However, the use of artificial intelligence (AI) technologies as mentioned above can be a double-edged sword It should be noted that they may face certain contentious issues, such as copyright, ethics, and balancing diverse and sustainable design (Khisamova, 2019; Zhu, 2021)
As of now, there is very little research literature on using AI to create packaging designs, and there is insufficient research evaluating the actual design effectiveness and time-saving efficiency for users of these AI technologies Therefore, we have conducted a study to measure whether using AI technology in product packaging design actually helps users save time or, in other words, to assess the acceptance of AI technology in the product packaging design process by users through the Time variable Here, we will focus on a single and recently popular technology from Microsoft, the Bing AI Image Generator From there, we can propose a hypothesis for this factor:
H1: In the process of product packaging design, users are willing to use the Bing
AI automated design technology because it can save time and effort
AI Design
In today's economic market, competition among businesses is always a crucial issue
Establishing a strong brand identity is crucial for businesses to differentiate themselves and succeed in the market This identity encompasses more than just visual elements like logos and colors; it includes creating consistent and recognizable elements that convey a brand's values and vision Packaging plays a vital role in brand identity, serving as a crucial touchpoint for customers to make judgments about a product By focusing on expressing unique brand characteristics, packaging can make products stand out and evoke strong brand recognition The aesthetic appeal of packaging, such as distinctive shapes and visually appealing designs, can positively influence brand evaluation, as demonstrated by successful brands like Coca-Cola and Absolut.
Heiz, or Perrier all have positive recognition (D., 1997) Indeed, packaging is the first aspect for recognizing, identifying, and differentiating brands and products in the increasingly crowded consumer goods market
When combined with the era considered the Internet age, businesses hope to integrate artificial intelligence (AI) into product packaging design to enhance brand recognition and differentiate themselves from competitors Wang and colleagues (2023) have observed that traditional packaging does not meet the increasingly complex needs of consumers in a growing and technologically advanced market Consequently, a concept of
"smart packaging design" has emerged as a key area to address this issue (Wang et al.,
2023) Furthermore, Wang has identified three key values of smart packaging design, enhancing safety and traceability, reflecting the increasing technology content of the product, as well as improving brand image and competitiveness in the market Indeed, according to Donal Norman's research (2005), he found a significant correlation between the function and emotional design of packaging (Norman, 2005) With the continuous development of artificial intelligence (AI), it can fully support designers in achieving remarkable results in customizing product packaging The application of AI technology in smart packaging design will shorten production time and enhance communication design methods (Milazzo & Libonati, 2022) This makes packaging increasingly smarter and more emotionally integrated, becoming a "living product" that consumers will always think of when needed Consider the following real-world examples from previous studies to see how famous companies apply AI technology to smart packaging design to enhance brand identity:
- In Hanna's study (2023), AI technology Midjouney Dall-E 3 was used to analyze and evaluate the effectiveness in brand recognition of famous brands such as Nike, McDonald's, and Mercedes Benz for 2025 (Hanna & Arts, 2023)
Figure 12: Applied Midjouney experiment for NIKE
Source: https://www.instagram.com/p/Cfbrr00oc5t/?img_index=1
Figure 13: Applied Midjourney experiment for Cinderella
Source: https://www.instagram.com/p/Cglc duMY9/
Figure 14: Applied Midourney experiment for Mercedes Ben
Source: Image from research of Hanna, 2022
Drawing inspiration from existing products of brands, the experimentation of using Midjouney in Nike's packaging design conducted by Jeffhandesign
(2022) demonstrated creativity and unique idea implementation (Jeffhandesign, 2022b) Through direct observation, users can easily recognize this as a Nike product thanks to the bold logo and creative application on their products Similarly, a promotional poster for Disney's famous movie, "Cinderella," references real-life imagery, helping viewers perceive the darkness in the character's life (Jeffhandesign, 2022a) Additionally, Hanna also experimented with this tool to convey the message of the future Mercedes Benz brand By using some prompts, she created easily recognizable images of the brand with a bold logo and modern, unique design style reflecting its message
Overall, the quality of these AI designs in product packaging design will impact customers' perception of the organization (Shams et al., 2015) Therefore, the quality of AI design seems to be a necessary condition for brand identification Previous studies have shown a positive correlation of this factor, but in reality, it is still in the experimental stage, so specific numbers to prove it have not been provided We also note there are very few studies examining whether users or consumers accept the use of this AI design and whether AI-designed packaging enhances brand recognition Therefore, in this research, we propose the following hypothesis:
H2: The application of AI technology in packaging design creates unique designs and high brand recognition, enhancing user identification and connection with the brand.
Perceived Attractiveness
From the analyses above, it can be seen that there is a relatively positive correlation between the effectiveness of product packaging design and brand recognition In the next section, we will delve deeper into the effectiveness of product packaging design through the concept of Perceived Attractiveness (PA) This concept refers to individuals' evaluations of an object, product, or even an idea as appealing and interesting from a psychological and marketing perspective (Kotler, 2018) It encompasses perceptions of attractiveness, beauty, and allure of an object or idea In the field of marketing communication, PA is often expressed through the combination of language and imagery (Jinglei Su, 2019) This has been confirmed that the level of language-image combination will impact consumer memory (Houston et al., 1987) and the persuasive effect of advertising (Areni et al., 1994) Previous studies have shown that information conveyed through product packaging, combining both language and imagery, can strongly influence customers' perception of attractiveness and enhance brand image in their minds In this way, product packaging contributes to creating and conveying meaning when described as a discourse (Dano, 1996) Indeed, according to Vitrac and Gaté (1993, p 128), product packaging allows the product to "speak out" while advertising refers to the product, "it
35 gives it a body, soul, and a reason for being." In this way, it provides narrative value to the product (Dano, 1996)
Therefore, to assess customers' perception of product attractiveness, designers in the product packaging process need to focus on the "Visual Stimulation" factor Applying visual stimulation can help highlight and capture customers' attention, thereby diverting them from the usual shopping process and disrupting their current selection mechanism In fact, the eye is an important factor in visual stimulation as it cannot help understand but instead record the sensation of a prominent and unexpected visual element entering the field of vision, such as unusual shapes and colors (D., 1973) Some loyal consumers of a specific brand sometimes do so just because of the better description of the product packaging, making them pay more attention For this reason, previous researchers have used semiotic analysis of consumer expectations and packaging expression to indicate a relationship Cavassilas (2006) discovered a semiotic system of the visual language of packaging and revealed forms of discourse such as types of signs (physical and figurative realms) and types of expressions (abstract, sensory, iconic, emotional, and affective) She also used examples from her research to show that sensory expression signifies yellow as dynamic and warm The choice and use of colors and shapes can influence customers' perception of product attractiveness According to Pantin-Sohier's study (2009), it allows brand managers to possess specialized knowledge in selecting and using colors and shapes (Pantin-Sohier, 2009) For example, round shapes and warm color palettes can express delicacy and femininity, while metallic tones or black can convey strength and masculinity This can be seen more clearly when explaining why Badoit mineral water is red, aiming to convey the feature of intense effervescence Recognizing this importance, image designers are putting great effort into analyzing the needs and preferences of consumers to create attractive and appealing designs, helping to build a special position for the brand in the minds of buyers However, combining vibrant graphics, unique shapes, or any attention- grabbing methods requires finesse (Rabinowitz, 2003) For example, the Coca-Cola bottle logo with its distinctive hourglass shape, or the design of Bordeaux wine bottles with
36 intricately decorated labels and elegant floral fonts, all evoke emotions and brand perception (Orth & Malkewitz, 2008) It can be said that beauty is a valuable direction, so consumer aesthetic preferences are often reflected through product design, even before the product's features
To meet the challenge, businesses have leveraged the rapid development of artificial intelligence (AI) to support the packaging design process In a study on consumer acceptance of AI-supported chatbots, they found that performance influences perceived attractiveness based on the theories of use and satisfaction (Maroufkhani et al., 2022) However, an important issue is the trustworthiness of the AI provider, affecting consumers' perceived attractiveness towards chatbots (Ameen et al., 2022) Nowadays, brands integrating AI technology into their branding strategy need to consider the credibility and reputation of the technology provider to enhance consumer attractiveness and acceptance Accumulatively, previous studies have shown that the perceived attractiveness of emerging
AI technologies hold potential in enhancing consumer acceptance within brand contexts, and performance expectations can influence such acceptance This study explores the application of Bing AI Image Generator design technology in product packaging, evaluating consumer acceptance through the variable "Perceived Attractiveness." We propose a hypothesis for this factor, recognizing the interplay between AI technology, product packaging design, and consumer perception.
H3: Perceived attractiveness increases the acceptance of emerging AI technologies in product packaging design influencing brand building.
Brand attitudes
In a highly competitive market, building a relationship between a brand and consumers (CBR) has become an important factor for marketing strategists Before studying the relationship between consumers and brands (CBR), the term "Brand Attitude" needs to be defined This is widely considered in evaluating consumer behavior (Fischer et al., 1994) "Attitude towards the brand" simply refers to the positive or negative thoughts that consumers have about your brand, based on their experiences and the ability to
Brand attitude refers to a network of interconnected beliefs and long-term opinions about a brand It plays a crucial role in predicting customers' reactions to marketing activities and their buying behavior A positive brand attitude often leads to higher prices, stronger customer-brand relationships, and increased loyalty Businesses can leverage AI to cultivate positive brand attitudes among consumers, who are increasingly receptive to AI-integrated products Furthermore, marketing strategies involving chatbots enhance communication and brand relationships through interaction, information, accessibility, entertainment, and customization.
38 these two concepts in the context of Artificial Intelligence (AI) delve deeper into the application of this technology in the field of product packaging design, and examine whether they are truly related to each other or not
"Brand Love" is the level of affection that consumers have for a specific brand, characterized by deep satisfaction and long-term commitment (Carroll & Ahuvia, 2006) When consumers have positive emotions toward a brand, it influences brand evaluation (Batra et al., 2012) According to Joshi & Garg (2020), brand love encompasses sustained commitment and satisfaction dependent on the brand exceeding expectations Although many studies have highlighted the unlimited benefits of Brand Love, it still remains a novel concept for many For instance, when consumers are dissatisfied with a product, they may generate negative word-of-mouth, which can have a detrimental impact on the business Brand love predicts emotions and can transmit emotions to others
Similarly, "Brand Trust" is the customer's belief in that brand Academically, "brand trust" refers to the consumer's trust in a product's ability to meet expectations to create satisfaction (Chinomona & Studies, 2016) When consumers trust that a product suits their needs, they will choose that brand to fulfill their requirements To build a strong bond with consumers, manufacturers need to focus on the level of trust in the brand This attachment often manifests through the brand image advertised across different product lines Research by Dumortier, Evans, Grebitus, & Martin (2017) has shown that trust in the brand is a decisive factor in consumers' purchasing decisions (Dumortier et al., 2017)
In general, according to previous studies, there is a positive interactive relationship between brand trust and brand love, falling within the scope of attitudes towards the brand (Albert & Merunka, 2013) Therefore, in the context of this study, we aim to investigate whether when using AI technology in product packaging design, such as the Bing AI Image Generator, consumers accept the images generated by AI and understand that this reflects trust and love for the brand This is a new and under-researched topic, combining the
39 features of AI into building customer-brand relationships, with limited previous literature on this issue Hence, the research team will propose a hypothesis to test this:
H4: High brand awareness will enhance consumer acceptance of using AI technology in product packaging design
RESEARCH MODEL
According to (Liu et al., 2021), accept AI technology is the public's acceptance of the willingness to accept the designs, the publications created by AI in general The acceptance of AI designs is based on customer satisfaction with the ability to combine colors and images, which is judged to be the right combination, without causing an unpleasant feeling of inhibition to the audience (Hanna & Arts, 2023)
Through the process of referencing previous research works, the research team has observed that the research topic related to artificial intelligence is still very new, with articles being published only since 2021 Along with that, studies on the application of AI in design have not yet made many discoveries, especially in the field of product packaging design Based on that observation, the research team has constructed a research model with
4 independent variables as described above, representing factors that may influence consumers' attitudes towards products packaging designed by AI
RESEARCH DESIGN
Research approach
To validate the research hypotheses, a quantitative research method involving a survey sample was employed Following data collection, the research team conducted data cleaning, filtering out erroneous or duplicate data Subsequently, the data was encoded into numerical form, with question titles assigned character designations (e.g., TM1 for Question 1 in the time variable) Once the dataset was complete, it was imported into SMARTPLS 4th version software for analysis The software was used to directly analyze the results displayed on its interface, requiring the dataset to be in csv file format.
Partial Least Squares Structural Equation Modeling (PLS-SEM)
In this study, the research team used the Partial Least Squares Structural Equation Modeling (PLS-SEM) method to evaluate the model on the SMARTPLS software The PLS-SEM model was applied to analyze data, providing a deep insight into the relationships between variables and helping to test research hypotheses that are exploratory or based on underdeveloped research foundations (Armstrong, 2014) PLS-SEM consists of 2 models: the structural model and the measurement model The Measurement Model, also known as the outer model, describes the relationships between latent variables and observed variables (indicators or items) The Structural Model, also known as the inner model, describes the relationships between latent variables The PLS-SEM model is essentially a path model that visually represents the relationships of research hypotheses and variables
The steps of data analysis are presented as follows:
(1) Draw measurement and structural models in SMARTPLS, including both observed and latent variables
(2) Set up analysis: In the model interface, configure parameters for the PLS Algorithm analysis, such as the number of iterations and the stopping criterion
(3) Run PLS Algorithm analysis: Perform analysis to estimate model parameters
(4) Bootstrap analysis: Conduct bootstrap analysis to test the stability and reliability of the estimates
(5) Evaluate the measurement model: Assess the reliability and validity of the measurement model through indices such as outer loadings and variable reliability
(6) Evaluate the structural model: Examine path coefficients and indirect effects to evaluate the relationships between latent variables
(7) Model validation: Use indices to evaluate the validity of the model.
Data collection
The research team conducted a non-probability sampling method, by this method the research team can actively select subjects to conduct surveys Data collection took place over a period of 15 days, from March 20th, 2024 to April 3rd, 2024 To observe statistical attitudes of AI users and consumers of products from the brands that the research team experimented with relatively and appropriately to the team's capabilities The survey focused on collecting feedback currently working in the two major cities of Vietnam, Hanoi and Ho Chi Minh City The research team also focused on sending questions mostly to individuals interested in the fields of design, photoshop The survey was sent online through various social media platforms and interest groups, forums related to the field of
43 design, suitable for the team's research theme, saving time and effort for the team while still ensuring the completeness of the questions The survey was conducted in both Vietnamese and English languages for convenient data collection
The questionnaire was designed through a scale of dependent and independent variables in the model presented earlier in the Research model section Based on the documents on the same topic that the research team has read and studied, the factors influencing consumer perception identified by the research team include factors: The composition material of the package (Ritnamkam, 2012), visual attractiveness (Orth, 2008)and brand attitude (Van Ooijen, 2017) are the factors that make up the overall appeal of the product Additionally, the research team also aims to observe the time efficiency of
AI in general and Bing AI in particular in image design, by surveying the average time users spend using Bing AI to create satisfactory artworks In the marketing 6P model (Armstrong, 2014), packaging is one of the powerful tools to attract and persuade target customers to believe and purchase the product The research team conducted in-depth interviews with consumers regarding packaging design, from which a general perception model was derived Based on the survey of consumer opinions on product packaging by (Chen, 2023), a survey questionnaire was developed The general perception model consists of 5 main factors, including: information transmission, brand recognition ability, functional usability, creative attractiveness, and user experience Specifically, these factors are
- Information Transmission: Consumers can easily recognize the effectiveness and characteristics of a product clearly and accurately through the design and images printed on the packaging
- Brand recognition: Consumers can identify which brand this product belongs to through the packaging design that accurately reflects the image and concept of that brand
- Functional usability: The packaging design must help users easily visualize the usage and effects of them, carrying practical significance
- Creative attractiveness: Users are visually attracted when exposed to the packaging design
- User experience: Design suitable for needs, aesthetics, and attractive to the target customer group emotionally
The questions we sent to the survey respondents are divided into different scales as presented in the table below:
Time Have you ever heard about Bing AI Image Creator (or AI create images like: Dall-e, Midjourney, Stable Diffusion ) before?
Have you tried using this technique? TM2
If yes, how much time do you spend drawing a perfect image with AI tools?
Do you feel AI helps save a tremendous amount of time to design? TM4
Do you feel AI helps save more time in design than the traditional way? TM5
AI Design In your opinion, do the above AI images help you better understand the brand/product?
By your visual experience, do you see a difference between normal design and AI design?
Are you feel posters designed using AI technology are impressing? AI3
Do the posters shown to you have a clear design idea? AI4
Can the creative AI technique be useful in enriching modern advertising? AI5
Can AI be able to design genius advertising solutions with a different vision?
Color on packaging Foster et al
Illustrations printed on the packaging PA3
How to arrange the layout PA4
I will consider buying these products that designed like this Cheng et al
I will recommend these products to those closest to me BA2
I believe the labels of logo written on packaging BA3
I will spread positive word-of-mouth about these packaging design BA4
These packaging made me want to use them BA5
Do you want to see more of these AI works and designs used in advertising?
Does seeing advertising posters using AI, call for meditation? et al (2021),
These packaging designs make me pay attention to the product being sold AC3
How do you feel about these brands from these packages? AC4
I believe this packaging has good quality AC5
These images help increase brand recognition AC6
According to (Sarstedt, 2019), to select the sample size for exploratory factor analysis (EFA), the ratio of observations / measured variables = 5:1, meaning that the sample size must be at least 5 times the total number of observed variables in the scale When designing the survey questionnaire, the research team used 26 observed variables in factor analysis, based on the ratio presented above, the minimum number of samples collected by the team needs to reach is: 26*5 = 130 observations, which means the team needs to collect at least valid responses from 130 survey forms During the data collection process, the research team received 249 responses, after using data cleaning methods, the team filtered out 235 valid responses, the effective response rate is 94%
The table below shows the results of the statistical demographic analysis of 235 valid respondents:
Table 2: Statistical analysis of demographics
Variable Value Frequency Number of subjects as % of total
Based on the questionnaire table above (see Table ), the gender ratio of female : male is 2:1, which is very suitable for the gender characteristics of people who have the habit of frequently buying products based on emotions (Bellenger, 1980) Considering the appropriateness in terms of age and usage purpose, the age group 16-22 is the group in the study and research stage, accounting for the highest proportion of respondents at 79.8% and 88.1% of individuals with a university education According to the survey sampling method in the study by Menon, D., & Shilpa, K (2023), this is the group of people with the most frequent need to use AI tools, so they have a certain understanding of similar products
SMARTPLS 4 provides three reliability evaluation indices or internal consistency indices: Cronbach's alpha, Composite reliability rho_a, Composite reliability rho_c (CR) According to (Marko Sarstedt, 2017), the focus should be on two indices: Cronbach's alpha and Composite reliability rho_c, where Cronbach's alpha assesses reliability too low, conversely, Composite reliability assesses reliability too high Therefore, when evaluating results, both indices should be presented, and the reasonable reliability value usually lies between Cronbach's alpha and Composite reliability
The formula for Cronbach’s alpha coefficient is calculated as follows:
● 𝑞 : the total number of questionnaires be distributed
● 𝑝𝑖 2 : the variance items within for the i item in order
● 𝑝𝑡 2 : the total score variance of all questionnaires
The Cronbach's α coefficient has a value between 0 and 1 A α value of greater than 0.6 is typically seen as indicative of internal consistency, whereas an α coefficient of 0.8
51 to 0.9 suggests a high level of scale dependability Reliability tests on the questionnaire's scale questions produced a value of 0.952, which was higher than 0.6 and suggested that the questionnaire's answers were extremely dependable
Composite reliability assessment is similar to Cronbach's alpha, represented by the following formula:
● ∑ 𝑛 𝑖 : the standardized loading of the 𝑖 𝑡ℎ item,
● 𝜗 𝑖 : the error variance associated with the 𝑖 𝑡ℎ item
The Cronbach's alpha coefficient, a measure of internal consistency reliability, ranges from 0 to 1 Values close to 0 indicate low reliability, while those close to 1 indicate high reliability For exploratory studies, a reliability coefficient between 0.6 and 0.7 is acceptable However, for more robust research, an optimal range of 0.7 to 0.9 is recommended.
- 0.9 (Jum Nunnally, 1994) If Composite reliability exceeds 0.95, multicollinearity may be present
The table below shows the results of the PLS-SEM algorithm analysis by the research team:
Table 3: Construct reliability and validity
The results show that all structural factors have good reliability when both Cronbach's alpha reliability coefficient and Composite reliability coefficient (rho_c) are greater than 0.7
RESULTS & FINDINGS
6.1 Structure model evaluation (PLS-SEM)
To evaluate the structural model, first need to consider the collinearity of the independent variables, in the model of the research group, the variables AI Design, Time, Perceived Attractive, Brand Attitudes play the role of independent variables, affecting the dependent variable defined as Acceptance AI Technology In cases of collinearity or multicollinearity, regression coefficients, p-values of significance of effects are biased leading to erroneous conclusions about relationships in the model According to (Hair et al.,
2011) To check whether the model is experiencing multicollinearity, it is necessary to monitor the threshold value of VIF (variance inflation factor) If VIF ≥ 5, there is a high likelihood of multicollinearity, leading to a serious impact on the model; in the case of 3 ≤ VIF < 5, it can be concluded that the model may be experiencing multicollinearity The safest threshold is when VIF < 3, at that threshold, we can continue analyzing the model without encountering multicollinearity issues The research team observed the VIF coefficient results of independent variables in the model run on SMARTPLS software as follows:
The results show that the independent variables in the model do not exhibit multicollinearity as all VIF values are below 3
Next, to draw conclusions about the research hypotheses, or in other words, to determine the significance of the relationships in the research model, we will use the results of the analysis of path coefficients In this section, we need to consider two aspects: (1) testing the statistical significance of the causal relationships and (2) evaluating the strength and direction of the causal relationships
In the step of testing the statistical significance of the causal relationships, the significance of a relationship (path coefficient) depends on its standard error obtained through bootstrapping in SMARTPLS 4 The bootstrap standard error allows us to calculate the t-value and p-value for all path coefficients in the structural model We can evaluate the statistical significance of the causal effect using the t-value or p-value, but usually, the evaluation is done based on the p-value, which is faster Typically, the critical values for two-tailed testing are 1.65 (significance level = 10%), 1.96 (significance level = 5%), and 2.57 (significance level = 1%) The commonly used significance level is 5% = 0.05, which is also the default level in SMARTPLS 4 If the Path coefficients result in a p-value smaller than 0.05, the effect is statistically significant Conversely, if the p-value is greater than 0.05, the effect is not statistically significant When a causal relationship is not statistically significant, we still keep that relationship in the model and conclude that it is not statistically significant, rather than removing the factor structure from the model
In the step of evaluating the strength and direction of the causal relationships, the default algorithm of SMARTPLS will output standardized path coefficients These coefficients range from -1 to +1 (values can be smaller/larger, but usually fall within the limits)
● A positive (+) path coefficient represents a positive relationship
● A negative (-) path coefficient represents a negative relationship
● A path coefficient close to +1 represents a strong positive relationship
● A path coefficient close to -1 represents a strong negative relationship
● A path coefficient closer to 0 represents a weak relationship
When multiple independent variables influence a dependent variable, to evaluate the strength of these independent variables' effects, we will rely on the magnitude of the path coefficients The magnitude is the absolute value of the path coefficient For example, if the independent variables have both positive and negative effects, we will take the absolute value of the path coefficient before comparing the strength To evaluate the causal relationships, we use the results of the Bootstrapping analysis The following table shows the results of the path coefficient testing
The formula for calculating T-statistics in the bootstrap model of SMARTPLS can be written mathematically as follows:
● Original Sample (O): This is the initial estimated value from the original sample
● Sample Mean (M): The average of the estimated values from all bootstrap samples
● Standard Deviation (STDEV): The standard deviation of the estimated values from the bootstrap samples
● T Statistics (|O/STDEV|): The T-statistics value is calculated by taking the absolute value of the ratio between the Original Sample and Standard Deviation
In this theory, P-Values indicate the impact relationship on independent paths, it is considered as the probability of obtaining a result at least equal to the observed value if hypothesis 0 (H0) is true, therefore if the P-values result shows a value ~ 0 or = 0, it means that the survey result is statistically significant At that time, when the T value is higher than the specified threshold (usually 1.96 for a significance level of 5%), it can be concluded that the parameter estimate is reliable
The table below shows the results of running the bootstrap model of the research group in SMARTPLS as follows:
The results of the bootstrapping model are mainly reflected in the results of the Original sample and P-values From this result table, the significance level of the t-test can be evaluated as reliable when this significance level is all lower than the comparison threshold of 0.05 At the same time, all effect coefficients are positive (+), thus all effect relationships in the model are in the same direction The order of effects from strongest to weakest on the variable Acceptance AI Tech is: Brand Attitudes (0.499) > Perceived Attractiveness (0.415) > Time (0.314) > AI Design (0.115)
Figure 16: : PLS-SEM algorithm result relationship model
Finally, we consider the R-squared representing the explanatory power of the independent variables on a dependent variable in the model In a model with how many variables acting as dependent, there will be as many R-squared coefficients According to Hair et al (2017), it is very difficult to provide a rule of thumb for accepting R-squared values, as this depends on the complexity of the model (few or many independent variables affecting the dependent variable, presence of mediating relationships ) and the research field Therefore, there is no convincing threshold to evaluate whether R-squared is achieved or not R-squared ranges from 0 to 1, approaching 1 means a high level of explanation for the dependent variable, approaching 0 means a low level of explanation for the dependent variable
We observe the results of the adjusted R-squared through the R-squared overview table:
The adjusted R-squared of Acceptance AI Tech is 0.857, thus the independent variables explain 85.7% of the variance of this variable
PROPOSED CONCLUSIONS AND SOLUTIONS
Conclusion and recommendation of research results
From the statistical results presented by the research group above, it is clear that respondents have a positive attitude and high agreement towards the Brand Attitudes and Perceived Attractive aspects brought by the design of AI, while the AI Design and Time aspects also have a fairly positive relationship with the acceptance of AI technology design Therefore, it can be concluded that AI image creation technology through the Bing AI tool brings an attractive feeling and clarifies the visual brand function on product packaging However, it can also be concluded that the optimal time factor in design and creativity in
AI design are not yet prominent This is the basis for the research group to propose some suggestions for practical application as presented below
Brand Recognition in AI-Generated Packaging Design: Based on the capabilities of Bing AI in designing image templates for surveys, it can be seen that AI has the ability to recognize the characteristics of a brand quite well and reflect that on product packaging design intelligently Bing AI Image Creator utilizes machine learning algorithms and image processing, this tool not only creates beautiful packaging design templates but also ensures that they accurately reflect the personality and values of the brand The application of this technique is not limited to creating attractive designs but also includes the ability to customize quickly according to specific customer requirements, helping businesses easily adapt to market trends and diverse consumer needs Therefore, brands can apply this feature to the packaging design process by providing detailed descriptions of their brand characteristics in the prompt, so that AI can generate images with accurate results, in line with the original idea The generated images can be tailored to specific product requirements, positioning the brand and target market At the same time, brands can leverage AI's packaging design to build a strong brand through product packaging Marketing strategies should focus on enhancing brand awareness and positive attitudes, as well as increasing product appeal to promote the acceptance of AI technology
Integrating AI technology into product design empowers creativity through Bing AI Image Creator, which generates captivating images and seamlessly transforms data into attractive packaging designs This integration enhances the design process, optimizing efficiency by reducing development time and costs in the research and design phases Moreover, it offers designers insights into current aesthetic trends and enables market demand prediction, providing a valuable resource for innovation.
Optimizing Design Time: Research has anticipated that the time spent on product design will have a positive relationship with the acceptance of AI technology, however, the results have shown that the duration of design does not necessarily have a significant impact on user and consumer acceptance of AI technology This may indicate that businesses can still apply AI to optimize packaging design processes, but instead of relying solely on AI capabilities, other optimization measures are still needed to shorten development time while maintaining the quality of the product design
Emotional Response to AI Design: The research results have revealed a fairly positive relationship between AI Design - users and consumers' emotions towards Bing AI designs, and the variable "Acceptance of AI Tech" has indicated a conclusion that, although
AI designs are evaluated as quite visually appealing, they still show that AI designs still have many flaws and evoke uncomfortable feelings for consumers Therefore, AI designs are only for reference ideas and require further adjustments to produce a complete design This result also shows that AI cannot completely replace the role of human designers in businesses
Discussion
The results after running the data using the SEM method through the SMARTPLS model have brought about noteworthy theoretical findings, contributing to a deeper understanding of the role of AI technology in the packaging design industry and the relationship between users and technology in today's commercial context
The study highlights the relationship between user and customer acceptance of Bing AI technology in packaging design This analysis enables a deeper understanding of how AI can be seamlessly integrated into the packaging design process, creating value for both users and businesses By elucidating this acceptance, the research provides valuable insights for effectively incorporating AI technology into packaging design, ensuring its widespread adoption and positive impact on user experiences.
Furthermore, this study has opened up opportunities to explore further the interaction between humans and AI technology in a commercial environment By analyzing user acceptance and adoption of AI technology in packaging design, we can better understand how this technology can be applied and optimized to create the best experience for end users
Additionally, the research sheds light on the increasingly important role of AI technology in the field of packaging design and advertising Understanding how this technology can be integrated into the design process and how it can interact with end users is crucial in creating the best products and shopping experiences
Finally, this study raises new questions and opens up avenues for further research Understanding more about the factors influencing user acceptance of AI technology in packaging design and the broader impacts of applying this technology in this field will be potential directions for future research
Current research has presented an adjusted model with reliability to assess the acceptance level of AI-enhanced design, thereby elucidating this phenomenon clearly The practical significance of these findings lies not only in expanding knowledge in the design field but also in providing crucial guidance for the development and application of related tools and technologies
Crucially, research advocates for the development of highly explainable, user-friendly, and accessible AI design tools This encompasses not only designing an intuitive user interface and providing clear instructions but also offering comprehensive support mechanisms to assist users in overcoming challenges and difficulties during the usage process.
Secondly, safeguarding privacy and intellectual property rights is also an issue that cannot be overlooked Developers need to concentrate on integrating security measures and controls to enhance user trust and minimize concerns regarding privacy and copyright risks
Thirdly, the user's interest and perceived value play a significant role in driving the use of AI design tools Emphasizing the benefits and utility of usage will stimulate curiosity and interest from users, thereby promoting the practical application of this technology in the design field
Lastly, providing comprehensive training and support for users not only helps them enhance knowledge and skills but also fosters confidence and positive behavioral intentions when using AI design tools This not only enhances work efficiency but also helps create a positive and innovative working environment
Limitation and furture research of the study
The research project "Measuring the acceptance of AI Bing technology in packaging design by users and AI customers" has achieved positive results, but there are still some limitations and further research is needed in the future to explore this issue more deeply
A key limitation is the narrow sample scope, which may not fully capture the diversity of users and customers who interact with AI Bing Expanding the sample size to include a more representative cross-section of users, perhaps through collaborations with businesses or user communities, would enhance the research's generalizability and provide a more comprehensive understanding of users' experiences.
Future research should investigate factors influencing user acceptance to provide a more comprehensive understanding These factors may include user experience, integration with practical needs, and privacy and security concerns By delving into these specific aspects, researchers can gain valuable insights into the factors that drive or hinder user adoption.
Moreover, although the results of this research show positive acceptance, they do not guarantee that this acceptance will continue to be strong in the future Follow-up studies are needed to assess the continuity of this acceptance under real conditions and in a dynamic market environment Especially, artificial intelligence (AI) is a prominent topic in recent times, especially artificial intelligence technologies in the field of image formation are also a novel topic Therefore, research works on this subject have not had many diverse topics for the research group to have a theoretical basis to clearly prove and provide a completely accurate overview In previous research works, the time factor in saving time is still considered an advantage of AI (Hanna & Arts, 2023) Therefore, when Time cannot yet have a strong positive impact on the acceptance of AI design, we still expect it to be the basis for us to continue developing research on equivalent topics in the future
Lastly, this research only focuses on measuring the acceptance of users and customers, without considering the broader impacts of using AI Bing technology in
64 packaging design, such as environmental effects and consumer protection Subsequent studies could broaden the scope to examine these impacts and find ways to optimize the use of AI technology in this field in a sustainable and beneficial manner for both users and the environment
In summary, the research project has achieved positive results, but further research is needed to delve deeper into the acceptance of users and customers towards AI Bing technology in packaging design, as well as to better understand the impacts and optimization potential of applying this technology in practice
APPENDIX A
Bing AI Image Creator: A product of AI that runs on Microsoft's DALL-E 3 artificial intelligence model, capable of generating images from prompt
Prompt: A reminder that describes the content that is inserted so that AI can learn and execute on user's request, here used in the sense of describing the content for the image that the research team wants Bing AI to create
Bing AI: Bing AI Image Creator
PLS-SEM: Partial Least Squares Structural Equation Modeling
APPENDIX B
List of research survey sample questionares
Time Have you ever heard about Bing AI Image Creator (or AI create images like: Dall-e,
Have you tried using this technique?
If yes, how much time do you spend drawing a perfect image with AI tools?
Do you feel AI helps save a tremendous amount of time to design?
Do you feel AI helps save more time in design than the traditional way?
AI Design In your opinion, do the above AI images help you better understand the brand/product?
By your visual experience, do you see a difference between normal design and AI design?
Are you feel posters designed using AI technology are impressing?
Do the posters shown to you have a clear design idea?
Can the creative AI technique be useful in enriching modern advertising?
Can AI be able to design genius advertising solutions with a different vision?
Color on packaging Packaging style Illustrations printed on the packaging How to arrange the layout
I will consider buying these products that designed like this
I will recommend these products to those closest to me
I believe the labels of logo written on packaging
I will spread positive word-of-mouth about these packaging design These packaging made me want to use them
Do you want to see more of these AI works and designs used in advertising?
Does seeing advertising posters using AI, call for meditation?
These packaging designs make me pay attention to the product being sold How do you feel about these brands from these packages?
I believe this packaging has good quality These images help increase brand recognition