1. Introduction

1.4. Research Framework

In order to gain a comprehensive understanding of the impact of AI on designers within enterprises and fill the research gap, we conducted interviews with 8 on-the-job design experts. Despite the existence of numerous quantitative research studies exploring the effects of AI on careers and work [4], there remains a lack of detailed field investigations and in-depth interviews that focus on the real experiences of designers in this field. Therefore, the purpose of this study was to observe the entire process of designers using AI tools, while documenting their behaviors and thoughts in the production of design briefs. By doing so, we aimed to gain insights into the boundaries of AI capabilities from the perspective of designers in enterprise settings.

Figure 1. Research Framework.

Figure 1. Research Framework.

The contributions of this study are as follows:

1. Identifies the primary characteristics that impact the use of AI technologies by designers during the project creation phase in the corporate work setting.

2. Thoroughly analyze the utilization of ChatGPT in the entire design summary creation process and enhance the original production framework.

3. Incorporate AI principles into the workflow, with the goal of enabling designers to successfully utilize AI technologies to enhance job productivity and quality.

2. Literature Review

2.1. Design Brief with Workflow

The design brief is a critical tool in project initiation, serving as a bridge between business and design [19]. It enables designers and clients to reach consensus and clarify project goals and expected outcomes [7]. The design brief's importance lies not only in providing a collaborative framework, including task assignments, resource organizing, tracking of key project milestones, and a step-by-step roadmap, but also in providing resources and processes that can be easily followed. Cross-functional collaboration is imperative in the process of product design. Parkman and Malkewitz (2019) further emphasize the significance of the design brief, viewing it as a product of knowledge cross-functional collaboration [20]. The Design Brief has three core functions: 1) Assessment: It serves as an evaluation tool to reflect the needs of stakeholders at the beginning of a project. 2) Agreement: It is an agreement between the client and the designer and serves as a critical validation tool that integrates business and innovation. 3) Documentation: As a working document for iterative development, the Design Brief guides the project continuously until completion.

The project brief is a process of organizing information by incorporating ideas from various stakeholders and writing it in a specific format. Although this format may vary from one organization or company to another, the logical process remains the same. To clarify and validate the intentions and roles of all parties involved in the design process, the design briefing should cover three stages: project initiation, project execution, and project completion. In these three stages, different stakeholders are responsible for writing the brief [18], as shown in Table 1. The client is responsible for defining the requirements and goals of the initial project brief, while the design team develops the brief based on this and proposes alternative solutions for constraints and costs, as well as validating the project implementation plan. Finally, after the completion of the design briefing, key content points are confirmed with the client, reducing deviations during formal project execution.

Table 1. Design brie's 3-Stage Process.

Table 1. Design brie's 3-Stage Process.

Stage	Responsible	Missions
Prepared	Designer, client	Define customer goals and expectations
Proceed	Designer	Create a plan list for the project design
Evaluation	Designer, client	Organize the final production resources and materials and confirm with customers.

To assess and evaluate the behavior patterns of designers when creating design briefs, it is necessary not only to understand the process but also to explore guiding principles and methods of template creation. Philip's widely accepted template creation method, presented in his 2004 work "Creating the Perfect Design Brief," summarizes seven key elements including "Project Overview and Background, Category Review, Target Audience Review, Company Portfolio, Business Objectives and Design Strategy, Project Scope, Timeline, and Budget, Research Data & Appendix"[21]. Similarly, Stone proposed a framework in 2010 consisting of ten elements including "Background summary, Overview, Driver, Audience, Competitors, Tone, Message, Visuals, Details, People" as headings for the structure and content. Regardless of which framework is adopted, the goal is to serve as a strategic manual to guide successful project outcomes. In addition to creating a content guideline, we also need an evaluation framework for design briefs. In this regard, we reference the Design Quality Criteria (DQC) framework as an evaluative model for the outputs of design briefs [9,22].

Table 2. Design Quality Criteria framework [22].

Table 2. Design Quality Criteria framework [22].

Dimension	Evaluation criteria
Strategy	Philosophy: the company's history, values, beliefs, vision, mission, and strategy
	Structure: The company's field, business model, and competitive advantage
	Innovation: The company's innovation areas and types
Content	Society: the needs and activities of individual or group consumers
	Environment: Requirements and expectations for environmental issues
	Feasibility: Expectations of economic efficiency
Performance	Process: Project budget and schedule
	Function: the nature of the deliverables, including the unique point of sales
	Expression: the sensory style and aesthetics of the product

The structure of creating a Design Brief may vary depending on the characteristics and complexity of the project [32]. Jones and Askland (2012) have identified seven key elements to ensure comprehensive production: 1) project overview; 2) market background and competitive analysis; 3) target audience evaluation; 4) company portfolio; 5) business goals and design strategy; 6) project scope; 7) research data and attachments [12]. These requirements clearly demonstrate the high demands on the designer's abilities when creating a Design Brief. The study also pointed out potential issues that may arise during the production process, such as a lack of real business logic, insufficient understanding of the design, and a lack of budget relevance. Designers often need to invest a significant amount of time in bridging the cognitive gap with clients [33]. GenAI, with its rich data sources and advanced neural networks, demonstrates impressive "knowledge transfer" capabilities [42]. This allows designers to better understand and meet project requirements, facilitating more effective design outputs, even when they face a lack of interdisciplinary knowledge in creating a Design Brief. In the following section, we will introduce AI tools and analyze their suitability for the role of an "assistant".

2.2. AI Tools in Design

In enterprise projects, the execution of design briefs is typically led by the product design manager. They possess extensive design experience and comprehensive design skills, enabling them to effectively manage design teams and control costs. To further explore how AI can assist these experts, it is essential to clarify the specific applications and potential of AI in the design field.

The application of GenAI based on large language models is expanding across various fields. According to McKinsey's analysis, over 70% of companies have deployed at least one type of artificial intelligence technology, and by 2030, the potential impact of artificial intelligence on global economic activity could be around $13 trillion. Within organizations, AI has a positive impact on management theory, including decision-making, knowledge management, customer service, human resources management, and administrative tasks [35]. In the field of design, GenAI's influence is continuously expanding, ranging from basic text and image generation to predicting user behavior and optimizing design decisions [27]. This provides designers with unprecedented efficiency improvements and innovation possibilities. The emergence of various AI tools is gradually changing the way user experience (UX) designers work and deeply influencing workflow within companies.

After preliminary investigation, we found that there are two main voices of designers' attitude toward AI in the company: one is "great, my work efficiency has been greatly improved, and I can focus on creativity and quickly generate expression"; The other is "my job has been replaced by AI, and I may lose my job." From a manager's point of view, they want their teams to be able to do their jobs more efficiently and cheaply, and when new productivity tools come out, they are willing to experiment, calling it a "revolution."

In Girling's (2017) research, he predicted that artificial intelligence would have a profound impact on the field of design [23]. He proposed that with advances in design technology, the learning curve would significantly decrease, allowing anyone to become a designer. Furthermore, he predicted a shift in the role of designers from creators to curators. These forecasts are highly far-sighted in today's context. We further explore how this transformation affects specific design practices. For instance, for designers focusing on 2D graphic design, they used to spend a substantial amount of time learning specialized 3D modeling software and setting parameters to create 3D models. However, with the development of artificial intelligence technology, this process has become simpler and more intuitive. Now, if designers can clearly articulate their ideas and convey them to AI through text prompts, they can quickly generate 3D models. This transition significantly lowers the barriers for designers to transition from 2D to 3D design, making the design process more convenient and efficient.

However, despite providing designers with new tools and possibilities, a study conducted by Bertão and Joo (2021) suggests that designers currently tend to view AI as an assistant to enhance workflow efficiency rather than a creative partner. This may be due to the limited ability of AI tools to understand and generate innovative designs [24]. Presently, AI's influence is primarily concentrated in fields such as UX/UI design, visual arts, industrial design, graphic design, and architecture. Nonetheless, this does not imply that other design fields such as architecture and fashion will not be impacted by AI. In fact, with the continuous development and application of AI technology, we can anticipate its influence spreading to more design domains. While we cannot list all possible impacts here, it is evident that most influences thus far are positive.

Table 3. The impact of GenAI on designers in design fields.

Table 3. The impact of GenAI on designers in design fields.

Domain	Impact on designers	Reference
UX/UI	Create scenarios, assist decision-making, improve efficiency	[24]
Fine Art	Inspire creativity, new expressions, and work with machines	[26] [30]
Graphic Design	Role transformation, efficiency improvement, innovation and experimentation, customization	[28] [31]
Industrial Design	Quickly generate and evaluate many options, improve innovation and efficiency	[29]

GenAI plays several common roles in the workflow. Firstly, it aids in cost forecasting [48,49,50]by utilizing precise data analysis and predictive models to assist enterprises in predicting and controlling project costs. Secondly, it optimizes project schedules [51,52] by employing algorithms and data analysis to ensure projects are completed on time. Thirdly, it enhances quality control [53] by monitoring and guaranteeing project quality, thereby reducing errors and defects. Furthermore, AI facilitates sustainability assessment [54,55,56] by assisting in environmental and sustainability evaluations, promoting green building and sustainable development. AI provides visualizations [57] that offer intuitive representations of data, enabling teams to better understand and analyze information.

When discussing the positive impact of artificial intelligence, we must also consider its controllability and boundaries. Once AI is implemented, there is no turning back, as computing power continues to evolve, rendering discussions on "boundary capabilities" short-sighted. With the constant emergence of new models, these boundaries may be redefined. However, we must remember that despite being a tool, AI's ultimate users are humans, and its development and application must be guided by human needs. Therefore, the continuous development of AI depends not only on technological advancement but also on human requirements and decision-making. This person-centered perspective is particularly crucial in the design industry.

Take ChatGPT as an example, which is a LLM-based. Despite its versatility, it fails to meet customization needs effectively and relies on user input and decision. The decision-making process of AI lacks transparency and poses various risks, such as inaccuracy, inherent bias, vision, privacy issues, and false information during output [4]. To address these issues, external plug-ins are currently used for correction, although achieving 100% accuracy is not possible. As demonstrated in Ferrara's (2023) research, the capabilities boundaries of AI and ChatGPT include fairness, bias and ethics, explainability and interpretability, auditability and accountability, controllability and security, and social impact. The usage of ChatGPT also entails challenges and risks related to data sourcing and filtering, as well as internal bias [34] Hence, designers need to be aware of these potential risks and learn how to effectively control and apply AI tools to avoid over-reliance.

Under the new wave of artificial intelligence, many work task processes are undergoing major adjustments and even disruption. A large body of literature [43,44,45,46,47] has demonstrated that AI can assist designers in their work in various ways. However, these studies often lack detailed descriptions of specific work scenarios and action processes, relying mainly on quantitative analysis of big data and literature inference. Moreover, few literatures directly discuss the feasibility and effect of using AI assistance in the production of Design briefs at the enterprise level. To fill this research gap and based on our hypothesis, we propose adding AI assistance to the production process of Design Brief, which can solve some problems in the manual production process. In the following sections, we will conduct in-depth research through interviews and experiments to test this hypothesis. We will communicate directly with the designer and conduct comprehensive interviews and records in the actual working scenarios. We expect that through this empirical research approach, we will be able to gain a deeper understanding of the application of AI in workflow.

3. Methodology

3.1. Experiment Design

Participants 

This study examines senior designers with over 5 years of experience, noted for their problem-solving skills and ability to secure clear design requirements through trust with clients. However, clear goals and requirements documents are often difficult to obtain at once. The requirements received by designers in enterprises are often ambiguous and manifested as "one-sentence requirements". The study targets senior designers with over 7 years of experience who can foresee project risks and minimize operational costs and risks. Eight such designers, leading various projects, were chosen for the test to ensure study reliability. To further enhance the fairness and objectivity of the research results, an independent expert was invited to anonymously evaluate the experimental output. The evaluation focused on three dimensions of "operability", "intelligibility", and "accuracy" to ensure impartiality.

Study Process 

The study examined designer experiences in creating design briefs across two phases. Initially, semi-structured interviews and tasks assessed designers' standard processes and challenges, focusing on RQ 1 and mapping key tasks and difficulties. Then, AI tools like ChatGPT were introduced to observe designers' performance with AI assistance, with further interviews exploring RQ2, 3, and 4, enabling a comparative analysis of outcomes with and without AI tools.

To maintain internal consistency and reduce bias, our qualitative study used a Within-Subject Design, accounting for environmental constraints and participant limits. This method enabled thorough examination and sampling of data. To mitigate the impact of order effects on the experimental findings, we divided the experimental procedure for each participant into two distinct stages. The initial stage aimed at alleviating psychological strain, consisting of a comprehensive interview lasting around 30 minutes. The second stage involved participants creating a Design Brief at regular intervals within the predetermined virtual project environment. We recorded the completion status and time of this stage, limiting it to one hour. Following the initial experiment, we invited participants again after a week, utilizing artificial intelligence methods to aid in the creation of the Design Brief. Throughout the exercise, we instructed participants to utilize the "Think Aloud" [36] to document their cognitive processes and decision-making rationale.

Figure 2. Experimental Process.

Figure 2. Experimental Process.

3.2. Experiment Setup

Questionnaire and Interview 

We conducted in-depth interviews to gather information on the preparation and implementation methods employed by designers in their work, to provide an empirical foundation for our research. The questionnaire was designed based on the Design Brief creation process and the fundamental components outlined in Chapter II. We acknowledge the existence of a disparity between theory and practice. However, by bridging this disparity, we can gain a deeper comprehension of the difficulties and possibilities that designers encounter.

The questionnaire was separated into two parts and seven sub-sections, which included basic information, briefing on the design process, problems, and suggestions for improvement. Each part seeks to collect precise information regarding how designers adjust and react to different project requirements and limitations.

In the first stage, "As Is" to understand the situation of designers making design brief.

-

Basic information: Background information of the designers was collected, including their experience, expertise, and types of previous projects.

-

Brief Design: Focuses on understanding how designers formulate and implement their design strategies based on the Design Brief.

-

Challenges: Specific problems and difficulties encountered by designers in implementing design strategies are explored.

-

Improvements: Specific suggestions and strategies from designers for improving the design process and overcoming challenges were collected.

The second stage, "To Be", is to understand how designers feel about using AI tools to create a design brief.

-

AI design brief experience: collect designers' subjective feelings on the use of AI in this project.

-

AI potential and challenges: Explore the potential and shortcomings of AI in workflows.

-

AI expectations: expectations for the future of AI-assisted workflows.

To achieve the main goal of this stage, which is to identify and analyze the differences and connections between real behavior and theory, a case-by-case approach is required. Understanding and documenting the real behavior and thinking of designers is necessary as each designer has their own unique work habits that are shaped by their experience and the specific needs and constraints of a project. Therefore, it is important to dedicate ourselves to this task.

User Experience map 

We utilize the user experience map as an assessment instrument to enhance the clear and concise representation of user behavior throughout usage [59]. Based on the prior literature review on "design brief production", the workflow of a designer can be categorized into three stages. During the prepare, the designer's primary focus is on engaging in effective communication with the client to establish the project goals and completing thorough market research to gain a comprehensive understanding of industry trends and user requirements. Given this knowledge, the designer will establish explicit design objectives and evaluate the project's time and resource demands. During the “Proceed” phase, the primary emphasis shifts towards generating design concepts, creating preliminary designs or prototypes, and engaging in design iteration and optimization through user feedback. At last, designers concentrate on enhancing the design specifics, compiling, and delivering comprehensive design papers, and collaboratively assessing the project outcomes with the client to ensure the successful conclusion of the project and offer following design assistance. The three stages mentioned above collectively form the entire process of design briefing production. This method guarantees the meticulous implementation and thoughtful examination of all project components. The table presented below delineates the fundamental framework of this experiment and serves as a methodical point of reference for the study.

Figure 3. Experience Map Framework based on design brief make [10].

Figure 3. Experience Map Framework based on design brief make [10].

Task Design 

To maintain control over the experimental conditions, we created a simulated project in which we established a sequence of pertinent tasks. To address the intricate nature of project management and the task of establishing project direction and objectives in real-world scenarios, we have endeavored to streamline this process while ensuring its practical applicability. This study specifically examines the essential tasks involved in the Design Brief process and integrates pertinent components of the UX map. Our meticulously crafted work scope encompasses the fundamental elements of the UX map, spanning from the beginning phase to the subsequent phase, with the objective of aiding designers in effectively visualizing the Design Brief process. Furthermore, the context in which these tasks are carried out serves as a reliable foundation for subsequent assessment of the designer's job performance.

Table 4. Task Design based on design brief.

Table 4. Task Design based on design brief.

Stage	No.	Task description	Execute
Prepared	1	study the company's history, values, beliefs, vision, mission, and strategies.	O
	2	Understand the company's field, business model, and competitive advantage.	O
	3	research the innovation field and type of the company.	X
	4	Conduct market surveys to understand consumer needs and activities.	O
	5	Economic feasibility assessment: Evaluate the cost and expected return of the design proposal.	O
Proceed	6	Develop project budgets and schedules.	O
	7	Clarify the function of the design scheme and ensure that it has a unique point of sales.	O
	8	Consider and design the sensory style and concept of the product.	O
Evaluation	9	Based on feedback and test results, improve design details.	X
	10	Prepare complete design documents, including drawings, specifications, etc.	X
	11	the benefit growth that can be brought after product production or optimization.	O
	12	Provide customers with follow-up design support and services.	O

As indicated in the table, the executable process in the virtual project has a total of 12 key tasks. The time required to create a Design Brief depends on the size of the project. For instance, in the case of an online game community platform, the initial draft is often developed by a team of 2~3 designers within a day's worth of work. Considering the nature of the experiment, we did not mandate designers to complete the entire Design brief. Instead, we chose specific tasks for testing purposes. In the “prepared” stage, we focused on tasks 1, 2, 4, and 5. In the “Proceed” stage, we examined tasks 6, 7, and 8. In the “Evaluation” stage, we assessed tasks 11 and 12. For simplified activities, like item 10, there is no requirement for the creation of a comprehensive design document, so it will be excluded during the testing phase.

ChatGPT 4.0 features a conversational interface, minimal learning requirements, and enables participants to effectively perform the job without the need for extra instruction or training [4]. This ensures the experiment progresses smoothly and guarantees the dependability of the collected data.

In the document evaluation step, a third-party expert evaluation was included, assessing operability, understandability, and accuracy across three dimensions. This project evaluation system is widely utilized throughout the firm. Operability pertains to assessing whether the actions outlined in the document can effectively accomplish the project objective. Understandability is employed to evaluate the ease with which the information in the document can be comprehended by the executor, thereby impacting the efficiency of communication. Accuracy ensures that all actions are aligned with the goal and contribute to its achievement.

Ultimately, we depict the phases, actions, interactions, emotions, challenges, and user experience diagrams of the testers involved in the production process of the Design brief. This allows for a clear and intuitive comprehension of the user behavior and process experience in the subsequent stages.

We studied designers' workflows and challenges in creating design briefs through interviews and questionnaires. We used user experience maps to visualize and compare workflows before and after AI tool implementation. Virtual tasks were created to control the experiment, and designer productivity was evaluated based on usability, comprehensibility, and precision.

4. Results

Table 5. Designer status.

4.1. The creation of the traditional design brief

We used in-depth interviews to understand the actual situation of designers dealing with design profiles, confirming the processes and elements of the previous literature review. The accuracy of the design profile is directly linked to the project schedule and cost. The following table lists the key points that the interviewed designers focused on and the factors that influenced the production of the design profile.

Table 6. Design brief production elements and executor concerns.

Table 6. Design brief production elements and executor concerns.

Key Elements	Description	Requirements
Clear goals	The correct goals can lay the foundation for the active promotion of the entire project.	Accuracy
	Vague goals can lead to confusion in team management.	Accuracy
Functional scope	The functional scope involves the cost estimation of the project, which will affect the allocation of manpower and time.	Accuracy
	Collaboration and scheduling affect all design functions.	Efficiency
	Evaluate the scope to judge the workload	Efficiency
Historical data and documents	Understanding the past information of the demand side can correct goals and avoid repeated mistakes.	Accuracy
	Have overall control over the development of the current project.	Accuracy
Stakeholder communication	Maintain consistency among all parties regarding project objectives.	Accuracy
	Reduce the loss of information transmission.	Efficiency
	Improve project execution efficiency.	Efficiency
	Actively communicate to ease conflicts of interest.	Accuracy
End users	Always keep its design goals in mind	Accuracy
	Improve the mining of correct requirements	Accuracy
Known constraints	Can strengthen control over team resources	Efficiency
	Correct task boundaries to prevent resource waste	Accuracy
Rapidly verifiable prototype	Used to quickly report or persuade stakeholders in non-design functions	Efficiency
	Improve the production efficiency of design introduction documents	Efficiency

The design brief's key components, including well-defined goals, functional boundaries, historical information, and documentation, significantly contribute to the project's success. Our data study revealed that accuracy and efficiency are the primary indicators of team performance in benefit-oriented organizations. The project's accuracy is crucial for ensuring it progresses in the correct path, and it contributes to 59% of the overall effectiveness of team management. Efficiency, which encompasses cost control, resource allocation, and teamwork, accounts for the remaining 41%. In the work, designers frequently encounter the necessity for repetitive revisions. The primary issues they confront are around ensuring the precision of goal setting and the judicious allocation of staff resources. Hence, designers require a method to enhance their productivity that is both precise and effective. Although it may not be flawless, if it reduces time and effort, it remains a viable choice to be considered.

We possess knowledge not just of the regular creation of design profiles, but also a preliminary comprehension of the assessment of existing AI technologies by designers. Feedback such as "The ChatGPT tool helped me clarify my needs," "It saved me a lot of time," and "ChatGPT's information needs to be handled carefully and occasionally shows inaccuracies and needs to be double-checked", showcases both the efficacy and constraints of the AI tool. Although most of the comments was favorable, there were a few reservations and worries. To gain a comprehensive understanding of how AI tools specifically affect designers' work, we will thoroughly examine the functioning and consequences of AI in practical scenarios during the second round of testing. The objective is to furnish helpful comments and recommendations for the subsequent advancement and utilization of AI solutions.

4.3. Execution of specified projects using artificial intelligence

Following the 1st testing, which occurred one week prior, we enlisted the participation of 8 designers for a subsequent round of testing. This time, we employed artificial intelligence techniques to execute and manage the identical work consistently. We conducted another round of testing using the prearranged material as a basis. The second testing procedure was performing the test followed by conducting the interview. Initially, in terms of the duration of the operation, this evaluation took 37 minutes, which is 48.6% longer than the previous round of testing, and the impact is noteworthy. The entire subjective experience has significantly enhanced, as evidenced by favorable feedback such as "the AI-generated report is exceptionally comprehensive", "it assists me in generating unexpected content" and "it efficiently searches for information based on my specific criteria." Additional recommendations include the necessity to re-validate the accuracy of the results, the requirement for manual correction due to the excessive length of the content, and the suggestion for a visual representation of the output.

During the same task evaluation, we saw a substantial enhancement in both operational experience and subjective perceptions with the assistance of AI. More precisely, designers that possess knowledge about the background of the target organization (client) see a 13.2% boost in their utilization of AI. Similarly, conducting market study leads to an 18.8% gain, while engaging in user and market research results in a 5.3% improvement. The economic feasibility assessment experienced a 16.7% increase. Budget and schedule management saw a 20.0% improvement. Feature clarification and optimization increased by 12.5%. Concept and style design improved by 7.1%. Project outcome estimates increased by 7.1%. Follow-up support and service also experienced a 12.5% increase. This data unequivocally illustrates the tangible advantages and prospective worth of AI tools in the daily tasks of designers.

Table 7. Comparison of two tests.

Table 7. Comparison of two tests.

No.	Task	Average score without using AI	Average score using AI	Difference
1	Understand the background of the target company (customer)	3.8	4.3	↑ 13.2%
2	Market analysis	3.2	3.8	↑ 18.8%
4	User and market research	3.8	4.0	↑ 5.3%
5	Economic feasibility assessment	3	3.5	↑ 16.7%
6	Budget and schedule management	2.5	3	↑ 20.0%
7	Function clarification and optimization	4	4.5	↑ 12.5%
8	Concept and style design	2.8	3	↑ 7.1%
11	Estimation of project results	2.8	3	↑ 7.1%
12	Follow-up support and services	4	4.5	↑ 12.5%

Figure 4. User experience map of actions and feedback from designers during the two tests.

Figure 4. User experience map of actions and feedback from designers during the two tests.

The main problem faced by the designers at the beginning of the project was the accuracy of the output of AI tools when processing a large amount of project data. For example, when working on a large project, ChatGPT's output may be distorted, requiring designers to double-check and correct it. Additionally, AI sometimes generates content, increasing the burden on designers to ensure accuracy. The output form of ChatGPT is relatively unitary, primarily producing literal output, which restricts designers' flexibility in addressing diverse needs. Moreover, the output content of AI tools demonstrates a certain degree of mechanical nature and limited judgment in different demand scenarios, lacking distinction and personalization based on complex stakeholder relationships. Designers also identified the lack of automation and prioritization of complex tasks in the AI tools. These problems partially limit the efficiency and effectiveness of applying AI tools in designers' daily work.

In the early stage of task testing, designers believed that AI tools could enhance their practical value by improving retrieval ability and explicitly checking the content. However, mid-term feedback revealed that increasing the form and personalization of the output content was the optimal direction for AI tools. For instance, in suggestion provision, ChatGPT can provide detailed feasibility assessments of solutions based on the requirements. Surprisingly, as depicted in the diagram, the content turned out to be much richer than what the designers had initially expected. Furthermore, designers identified automation as a crucial capability they anticipated from AI tools.

We presented two iterations of design brief reports to external experts for assessment, focusing on three criteria: usability, comprehensibility, and precision. The evaluation results were generally consistent with predictions, with improvements observed in the operability and understandability of the two rounds of reports. However, there was a minor drop in accuracy, although the total difference was not significant. AI provides a more thorough study of the alignment between tasks and requirements. Illustrating the case of "Scheduling task of human workload," ChatGPT promptly provided three distinct possibilities with the assistance of the operator and evaluated their advantages and disadvantages. Conversely, human evaluation predominantly depends on designers' own experience to make conclusions and judgments, which is susceptible to hasty circumstances. Given the constraints of the testing period, the AI delivered more extensive responses in this area, resulting in good ratings for usability and comprehensibility. Nevertheless, ChatGPT's accuracy falls short compared to human evaluation due to the verbosity of the generated material and the potential distortion of information in the absence of a network connection.

Table 8. Design brief production elements and executor concerns.

Table 8. Design brief production elements and executor concerns.

Evaluation	1st group	2nd group	Difference
Operability	4.5	4.7	↑ 4.44%
Understandability	4.2	4.4	↑ 4.76%
Accuracy	4.8	4.6	↓ 4.17%
Time	72min	37min	↑ 49%

5. Discussion

5.2. Design with AI Thinking

AI Thinking in Enterprise Workflows 

In traditional design methodologies, designers frequently encounter obstacles involving data collection, requirement validation, and the integration and organization of vast amounts of information. Obtaining accurate data can be time-consuming and necessitate considerable effort [32,33]. Nonetheless, due to project time constraints, the resulting documentation may occasionally encounter execution difficulties, leading to continuous revisions. This not only escalates the workload but also results in substantial cost losses for the project team. To address these issues, Jones, W. M., & Askland, H. H. (2012) propose an open and flexible approach to the production of a design brief. [12]

For business managers, their primary concern is the data related to production costs. Any tool that can improve efficiency and optimize production processes deserves their attention. As a result, nearly all production processes, not merely in the design brief, hold the potential to benefit from the integration of such tools. Drawing upon current trends, AI transcends being merely an efficiency tool and necessitates standardization and profound integration into the thinking of the production process.

For enterprises, the following are the findings of this study:

AI Workflow Management 

-

Cost management: In the realm of task execution engineering, we observe that numerous processes are predefined, necessitating the automation of tasks, optimization of resource allocation, and reduction of human error, particularly in data processing

-

Risk management:During the testing phase, ChatGPT surprised us by providing suggestions related to market risks and personnel allocation constraints based on the existing data. For example, it advised that due to the complexity of the task and the current personnel allocation, there was a potential risk of delay in completing the task within the stipulated time. This suggestion was later confirmed by evaluation from design experts.

-

Anticipation management:The analysis of client needs is a crucial part of project execution. AI technology enables businesses to better understand and meet customer needs and expectations. For example, by analyzing market data and user feedback, companies can predict market trends and customer demands, helping them provide more personalized products or services.

Collaboration 

-

Human-computer collaboration:The involvement of AI prompts us to reconsider the collaboration between humans and computers. Computers excel at handling large amounts of data and repetitive tasks, while designers possess unique advantages in creativity, strategy, and emotional interaction. Effective human-computer collaboration not only enhances work efficiency but also allows designers to focus more on innovation and high-level decision-making.

-

Post collaboration:The differences in knowledge systems among various functions necessitate a significant time cost for effective communication during collaboration. However, the characteristics of AI can overcome the cognitive gap between functions. For instance, by utilizing AI analysis, users can predict project timelines and resource requirements with greater accuracy, thus enabling better task allocation and team management.

The Impact of AI on Designers 

In the previous section's analysis, we clearly observed the significant impact of AI tools on improving the efficiency of designers. Human-computer interaction (HCI) primarily focuses on the interaction process between humans and information, and with the intervention of AI, many tasks related to information exchange have been automated. This has led to the emergence of the advanced concept of "Human-Computer AI Interaction" (HCAI) [40]. In this context, AI is not just a simple tool; it represents a completely new way of thinking. Mastering and skillfully applying this AI thinking will be crucial for future product development and project planning. Through interviews and testing tasks, we identified five points that highlight the views of designers on AI tools when using ChatGPT.

Searching: Traditionally, a significant amount of time is required to acquire background knowledge when establishing an information framework. However, ChatGPT's generative capabilities enable direct management of relevant content, acting as both a search and retrieval tool. For instance, by inputting a keyword, the AI can rapidly provide a substantial amount of related information, facilitating designers in swiftly gathering and organizing data.

Verification: AI can process vast amounts of data and information efficiently, offering immediate multiple-choice feedback to the operator. This feedback aids the designer in identifying and rectifying potential issues early on, minimizing the need for extensive revisions in the future.

Analysis:AI tools can analyze extensive user data, improving designers' comprehension of customer requirements.

Communication: To enhance communication among designers, team members, clients, and stakeholders, it is essential to explore new frontiers of knowledge.

Decision Making: AI can offer designers data-driven recommendations to support decision-making. Through extensive data analysis, AI can predict the potential outcome of specific design decisions, enabling designers to make more informed choices.

The feedback above indicates that AI technology has improved efficiency for designers through fast information retrieval, immediate feedback, in-depth data analysis, and decision-making recommendations. Additionally, it has facilitated communication between designers, teams, and clients. However, it also presents challenges, such as concerns about the authenticity of information, copyright issues, and potential job displacement, leading to professional anxieties.

Error Messages: Designers generally express concerns about the credibility of AI information. These inaccuracies may originate from biased training data, algorithm limitations, or other factors. Therefore, designers should maintain critical thinking when using AI tools and verify and review the information provided.

Content Copyright: We used ChatGPT as our primary research tool for this study. However, copyright issues pose a significant challenge when it comes to multimedia content such as images, videos, or audio. To address these issues effectively, the industry needs to establish robust standards that clearly define the copyright ownership and usage rights of AI-generated content. Additionally, internal company policies and guidelines should be developed to ensure the proper handling of copyright issues and mitigate potential legal risks when using AI tools.

Expert Status: During our initial research, we found that many companies are reevaluating their workflow processes, especially in the design and software development departments, which have been significantly impacted by AI. For instance, in the field of graphic design, approximately 50% of the work is now being replaced by graphic AI tools like MidJourney and Stable Diffusion. Similarly, in software development, there are tools such as co-pilot that help developers. While these AI tools enhance productivity, they may also result in the elimination of certain positions, leading to career concerns among employees.

The application of AI is not limited to the production of concise designs. Instead, it can play a role in the entire workflow, aiding businesses in achieving their goals and enabling employees to complete their tasks more effectively. As shown in the diagram below, we integrate AI thinking into key processes like formulating business strategies, generating document content, and making design decisions. We also specify the specific tasks that AI can undertake at each stage.

Figure 5. AI Thinking in Workflow.

Figure 5. AI Thinking in Workflow.

6. Conclusion

References

1. Manyika, J., et al. Jobs Lost, Jobs Gained: What the Future of Work Will Mean for Jobs, Skills, and Wages. McKinsey Global Institute, 2023.
2. WEF. The Future of Jobs Report 2023. World Economic Forum, 2023.
3. Al-Sa’di, A.; Miller, D. Exploring the Impact of Artificial Intelligence language model ChatGPT on the User Experience. Int. J. Technol. Innov. Manag. 2023, 3(1), 1–8. [Google Scholar] [CrossRef]
4. Eloundou, T.; Manning, S.; Mishkin, P.; Rock, D. GPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Models. 2023. [Google Scholar] [CrossRef]
5. Hassenzahl, M.; Tractinsky, N. User experience - a research agenda. Behav. Inf. Technol. 2006, 25(2), 91–97. [Google Scholar] [CrossRef]
6. Bertão, R. A.; De Souza Brum, A. L.; Joo, J. The design brief as a pivotal tool: A study of Centro Brasil Design’s practices to promote design. Des. J. 2023, 26(2), 188–209. [Google Scholar] [CrossRef]
7. Paton, B.; Dorst, K. Briefing and Reframing: A Situated Practice. Des. Stud. 2011, 32(6), 573–587. [Google Scholar] [CrossRef]
8. Jevnaker, B. H. Vita Activa: On Relationships between Design(Ers) and Business. Des. Issues 2005, 21(3), 25–48. [Google Scholar] [CrossRef]
9. Petersen, S. O.; Phillips, P. C. Inspiring Design-Informed by Metrics. Des. Manag. Rev. 2011, 22(2), 62–71. [Google Scholar] [CrossRef]
10. Petersen, S. E. DEVELOPING AN INSPIRATIONAL DESIGN BRIEF. In Proceedings of DESIGN 2010, the 11th International Design Conference, Dubrovnik, Croatia; 2010; pp. 1173–1182. [Google Scholar]
11. Al-Sa’di, A.; Miller, D. Exploring the Impact of Artificial Intelligence language model ChatGPT on the User Experience. Int. J. Technol. Innov. Manag. 2023, 3, 1–8. [Google Scholar] [CrossRef]
12. Jones, W. M.; Askland, H. H. Design Briefs: Is there a Standard? In Proceedings of the 14th International Conference on Engineering & Product Design Education (E&PDE12) Design Education for Future Wellbeing; 2012; pp. 115–120. [Google Scholar]
13. Stickdorn, M.; Schneider, J. This is Service Design Thinking: Basics – Tools – Cases. 2012, pp:151-154.
14. Tang, V.; Liu, B.; Kellam, K.; Otto, K.; Seering, W. Comparison of success factors. ICED05 Melbourn 2005.
15. Cooper, R. G. Product Leadership - Creating and Launching Superior New Products. Perseus Books: Cambridge MA, 1998.
16. Petersen, S. Design Quantification – Design Concept Argumentation as related to product performance Metrics. Stanford University, 2009.
17. Berni, A.; Borgianni, Y. FROM THE DEFINITION OF USER EXPERIENCE TO A FRAMEWORK TO CLASSIFY ITS APPLICATIONS IN DESIGN. Proc. Des. Soc. 2021, 1, 1627–1636. [Google Scholar] [CrossRef]
18. Blyth, A.; Worthington, J. Managing the Brief for Better Design. Routledge eBooks, 2010. [CrossRef]
19. Jevnaker, B. H. Vita Activa: On Relationships Between Design(ers) and Business. Des. Issues 2005, 21(3), 25–48. [Google Scholar] [CrossRef]
20. Parkman, I. D.; Malkewitz, K. Design briefs in design-driven new product development. J. Des. Bus. Soc. 2019, 5(1), 35–58. [Google Scholar] [CrossRef] [PubMed]
21. Stone, T. L. Managing the Design Process: Implementing Design: An Essential Manual for the Working Designer. 2010.
22. Petersen, S. O.; Joo, J. Inspirational Design Briefing. In John Wiley & Sons, Inc. eBooks, 2015; pp. 13–26. [CrossRef]
23. Girling, R. AI and the future of design: What will the designer of 2025 look like? O’Reilly Media, 2017. Available online: https://www.artefactgroup.com/ideas/ai_design_2025/ (accessed on [date]).
24. Bertão, R. C.; Joo, J. Artificial intelligence in UX/UI design: a survey on current adoption and [future] practices. In Blucher Design Proceedings, 2021. [CrossRef]
25. Dellermann, D.; Calma, A.; Lipusch, N.; Weber, T.; Weigel, S.; Ebel, P. The Future of Human-AI Collaboration: A Taxonomy of Design Knowledge for Hybrid Intelligence Systems. 2019. [Google Scholar] [CrossRef]
26. DiPaola, S.; Gabora, L.; McCaig, G. Informing artificial intelligence generative techniques using cognitive theories of human creativity. Procedia Comput. Sci. 2018, 145, 158–168. [Google Scholar] [CrossRef]
27. Zhu, Q.; Luo, J. Generative Transformers for Design Concept Generation. J. Comput. Inf. Sci. Eng. 2022, 1–61. [Google Scholar] [CrossRef]
28. Forrester. Generative AI Needs Design. Forbes, 2023. Available online: https://www.forbes.com/sites/forrester/2023/02/17/generative-ai-needs-design/?sh=71727c6e6575 (accessed on [date]).
29. RedBlink. Generative AI in Product Design & Development - Benefits & Tips. 2023. Available online: https://redblink.com/generative-ai-in-product-design/ (accessed on [date]).
30. Mukherjee Chakraborty, M. How Generative AI is Disrupting Traditional Art and Design Industries. Tech Blogger, 2023. Available online: https://contenteratechspace.com/how-generative-ai-is-disrupting-traditional-art-and-design-industries/ (accessed on [date]).
31. Shevde, R. How Generative AI is affecting Graphic Design in 2023. 2023. Available online: https://designwizard.com/blog/how-generative-ai-is-affecting-graphic-design/ (accessed on [date]).
32. Phillips, P. L. Creating the Perfect Design Brief: How to Manage Design for Strategic Advantage. Allworth Press: New York, 2012.
33. Dewulf, K. Integrating Sustainability in the Strategic Stage of an Innovation Process: A Design Brief Perspective. In IntechOpen eBooks, 2020. [CrossRef]
34. Ferrara, E. Should ChatGPT be Biased? Challenges and Risks of Bias in Large Language Models. 2023. [CrossRef]
35. Korzynski, P.; Mazurek, G.; Altmann, A.; Ejdys, J.; Kazlauskaite, R.; Paliszkiewicz, J.; Wach, K.; Ziemba, E. Generative artificial intelligence as a new context for management theories: analysis of ChatGPT. Cent. Eur. Manag. J. 2023, 31(1), 3–13. [Google Scholar] [CrossRef]
36. Birns, J. H.; Joffre, K. A.; Leclerc, J. F.; Paulsen, C. A. Getting the whole picture: Collecting usability data using two methods––Concurrent think aloud and retrospective probing. In Proceedings of UPA Conference, 2002; pp. 8-12.
37. Olga, A.; Saini, A.; Zapata, G.; Searsmith, D.; Cope, B.; Kalantzis, M.; Kastani, N. P. Generative AI: Implications and Applications for Education. arXiv preprint arXiv:2305.07605, 2023.
38. Jeong, C. A Study on the Implementation of Generative AI Services Using an Enterprise Data-Based LLM Application Architecture. arXiv preprint arXiv:2309.01105, 2023.
39. Singh, A.; Jia, T.; Nalagatla, V. Generative AI Enabled Conversational Chatbot for Drilling and Production Analytics. In Abu Dhabi International Petroleum Exhibition and Conference, 2023; p. D021S065R002. SPE.
40. Fui-Hoon Nah, F.; Zheng, R.; Cai, J.; Siau, K.; Chen, L. Generative AI and ChatGPT: Applications, challenges, and AI-human collaboration. J. Inf. Technol. Case Appl. Res. 2023, 25(3), 277–304. [Google Scholar] [CrossRef]
41. Ghimire, P.; Kim, K.; Acharya, M. Generative AI in the Construction Industry: Opportunities & Challenges. arXiv preprint arXiv:2310.04427, 2023.
42. Ritala, P.; Ruokonen, M.; Ramaul, L. Transforming boundaries: how does ChatGPT change knowledge work? J. Bus. Strateg. 2023. [CrossRef]
43. MayankAgarwal, J. J. Barroso, T. Chakraborti, E. M. Dow, K. Fadnis, B. Godoy, M. Pallan, and K. Talamadupula. Project clai: Instrumenting the command line as a new environment for ai agents. arXiv preprint arXiv:2002.00762, 2020. arXiv:2002.00762, 2020.
44. Ryan Louie, A. Coenen, C. Z. Huang, M. Terry, and C. J. Cai. Novice-AI music co-creation via AI-steering tools for deep generative models. In Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems, 2020; pp. 1–13.
45. Jiao Sun, Q. Vera Liao, M. Muller, M. Agarwal, S. Houde, K. Talamadupula, and J. D. Weisz. Investigating Explainability of Generative AI for Code through Scenario-based Design. In 27th International Conference on Intelligent User Interfaces, 2022; pp. 212–228.
46. JustinDWeisz, M. Muller, S. Houde, J. Richards, S. I. Ross, F. Martinez, M. Agarwal, and K. Talamadupula. Perfection Not Required? Human-AI Partnerships in Code Translation. In 26th International Conference on Intelligent User Interfaces, 2021; pp. 402–412.
47. JustinDWeisz, M. Muller, S. I. Ross, F. Martinez, S. Houde, M. Agarwal, K. Talamadupula, and J. T. Richards. Better together? an evaluation of ai-supported code translation. In 27th International Conference on Intelligent User Interfaces, 2022; pp. 369–391.
48. Ghimire, P.; Pokharel, S.; Kim, K.; Barutha, P. MACHINE LEARNING-BASED PREDICTION MODELS FOR BUDGET FORECAST IN CAPITAL CONSTRUCTION. 2023.
49. Baduge, S. K., et al. Artificial intelligence and smart vision for building and construction 4.0: Machine and deep learning methods and applications. Autom. Constr. 2022, 141, 104440. [CrossRef]
50. Mahmoodzadeh, A.; Nejati, H. R.; Mohammadi, M. Optimized machine learning modeling for predicting the construction cost and duration of tunneling projects. Autom. Constr. 2022, 139, 104305. [Google Scholar] [CrossRef]
51. Abioye, S. O., et al. Artificial intelligence in the construction industry: A review of present status, opportunities and future challenges. J. Build. Eng. 2021, 44, 103299. [CrossRef]
52. Xu, Y.; Zhou, Y.; Sekula, P.; Ding, L. Machine learning in construction: From shallow to deep learning. Dev. Built Environ. 2021, 6, 100045. [Google Scholar] [CrossRef]
53. Saravanan, V.; Pourhomayoun, M.; Mazari, M. A Proposed Method to Improve Highway Construction Quality Using Machine Learning. In 2018 International Conference on Computational Science and Computational Intelligence (CSCI), 2018; pp. 1218–1221. [CrossRef]
54. Kar, A. K.; Choudhary, S. K.; Singh, V. K. How can artificial intelligence impact sustainability: A systematic literature review. J. Clean. Prod. 2022, 376, 134120. [Google Scholar] [CrossRef]
55. Debrah, C.; Chan, A. P. C.; Darko, A. Artificial intelligence in green building. Autom. Constr. 2022, 137, 104192. [Google Scholar] [CrossRef]
56. Fathi, S.; Srinivasan, R.; Fenner, A.; Fathi, S. Machine learning applications in urban building energy performance forecasting: A systematic review. Renew. Sustain. Energy Rev. 2020, 133, 110287. [Google Scholar] [CrossRef]
57. Seo, J.; Park, H.; Choo, S. Inference of Drawing Elements and Space Usage on Architectural Drawings Using Semantic Segmentation. Appl. Sci. 2020, 10(20), Art. no. 20. [CrossRef]
58. Kim, H.; Sefcik, J. S.; Bradway, C. Characteristics of qualitative descriptive studies: A systematic review. Res. Nurs. Health 2017, 40(1), 23–42. [Google Scholar] [CrossRef]
59. Lallemand, C.; Lauret, J.; Drouet, L. Physical Journey Maps: Staging Users’ Experiences to Increase Stakeholders’ Empathy towards Users. In CHI Conference on Human Factors in Computing Systems Extended Abstracts, 2022; pp. 1-7.
60. Li, J.; Cao, H.; Lin, L.; Hou, Y.; Zhu, R.; El Ali, A. B. D. A. L. L. A. H. User Experience Design Professionals’ Perceptions of Generative Artificial Intelligence.