Exploringthe Potential of Artificial Intelligence as a Tool for Architectural Design: A Perception Study Using Gaudí’sWorks
Round 1
Reviewer 1 Report
This is a topic of great interest to the field of architecture as it is framed in what some authors have called the fourth industrial revolution driven by the exponential increase in computing power. But the way the article is presented raises numerous questions that should be resolved before being published:
1.- The focus of the article should be artificial intelligence and not the case through which its potential is studied. It is therefore recommended that the title of the article be changed.
2.- There is little depth in the approach to the key concepts of the study, only a few references to studies are presented, but without allowing the reader a clear understanding of the interpretation given to them in this study. Furthermore, the approach is clearly ocular-centric when architecture is multi-sensory. This issue should have been addressed in order to understand how artificial intelligence deals with this reality. This is a major omission. The paper needs the theoretical framework section to be well developed.
3.- There is terminology associated with the functioning of artificial intelligence that should be explained in the article.
4.- Methods: The paper presents some relations between quantitative measurements and sensorial evaluations that are not justified. It is recommended to introduce some quality mechanisms to ensure the validity of outcomes.
It is uncertain whether the applied methodology will be able to grant conclusive results or establish a clear relationship between quantitative measurements and emotional evaluations.
On the other hand, a survey using the Likert scale measures attitudes in an explicit way, but it may rise an issue: What about the implicit measures in order to avoid existing stereotypes or prejudices? There is a considerable risk to fall into reductionism if there is a lack of investigation into the social class cultural influence in emotional evaluation since culture teaches how to feel. Therefore, this lack does not allow for knowing if this method would produce the expected results.
There is no information on how many people from each age group, and on how many males and females participated in surveys.
5.- It introduces matters that are repeated throughout the text but not dealt with in-depth such as the question of ethics in the use of these technologies.
Author Response
Response to reviewers:
First and foremost, I want to express my deep gratitude for your time, insights, and constructive critiques. They have greatly aided in refining this manuscript and have significantly contributed to my own understanding and growth in this field. Your expert reviews are immensely appreciated, and I am eager to address your concerns.
Below, I have responded to your comments and made relevant amendments to the manuscript:
#1 reviewer
Concern #1: The focus of the article should be artificial intelligence and not the case through which its potential is studied. It is therefore recommended that the title of the article be changed.
Author response: Thank you for your suggestion. We have changed the title of the article to better reflect the focus on artificial intelligence. ¨Exploring AI's Potential in Architectural Visualization: A Perception Study Using Gaudí's Works¨
Concern #2: There is little depth in the approach to the key concepts of the study, only a few references to studies are presented, but without allowing the reader a clear understanding of the interpretation given to them in this study. Furthermore, the approach is clearly ocular-centric when architecture is multi-sensory. This issue should have been addressed in order to understand how artificial intelligence deals with this reality. This is a major omission. The paper needs the theoretical framework section to be well developed.
Author response: We appreciate your feedback. Following your advice,We have entirely rewritten the second and third sections of the literature review in our paper. The key concepts have been rearticulated, and a thorough discussion about the importance of the ocular-centric viewpoint in architecture has been integrated. I have woven the discussion on ocular-centricity throughout the text to explain how artificial intelligence contends with this reality.
The revisions in response to these issues can be found in lines 90 to 177 of the manuscript.
Concern #3: There is terminology associated with the functioning of artificial intelligence that should be explained in the article.
Author response: Thank you for noting this. We have included explanations of terminology associated with the functioning of artificial intelligence in the article.
The revisions in response to these issues can be found in lines20，168，211，221 of the manuscript.
Concern #4: The paper presents some relations between quantitative measurements and sensorial evaluations that are not justified. It is recommended to introduce some quality mechanisms to ensure the validity of outcomes. It is uncertain whether the applied methodology will be able to grant conclusive results or establish a clear relationship between quantitative measurements and emotional evaluations. On the other hand, a survey using the Likert scale measures attitudes in an explicit way, but it may rise an issue: What about the implicit measures in order to avoid existing stereotypes or prejudices? There is a considerable risk to fall into reductionism if there is a lack of investigation into the social class cultural influence in emotional evaluation since culture teaches how to feel. Therefore, this lack does not allow for knowing if this method would produce the expected results. There is no information on how many people from each age group, and on how many males and females participated in surveys.
Author response: We appreciate your feedback and have added quality mechanisms to ensure the validity of our outcomes.
1- In response to your first concern, we have clarified the relationship between quantitative measurements and sensory evaluations and introduced certain quality mechanisms to ensure the validity of the outcomes. We have clearly described the methodology applied, how it can provide conclusive results, and establish a clear relationship between quantitative measurements and emotional evaluations. (can be found in lines 278 to 308)
2-Regarding the potential issue you highlighted with the use of the Likert scale, we concur that it indeed represents a limitation. Despite our attempt to sidestep existing stereotypes or prejudices through the data collected from our questionnaire, we acknowledge that it is not completely possible to evade the influence of social class culture on emotional evaluation, as culture indeed instructs us on how to feel. We will delineate this limitation explicitly in the discussion section of the paper. (Can be found in lines 444 to 456)
3-Concerning the information on the number of participants and the distribution across different age groups and genders, we have supplemented the necessary data in the manuscript. We concur that the absence of students from the architecture department in our research is regrettable. However, we do not see a necessity to differentiate between males and females, as it is not the aim of our research. Our study primarily explores the possibilities of AI in architectural design and the acceptance level of AI-generated designs by individuals, which is not influenced by gender. (Can be found in lines 238 to 256, lines 457 to 463)
Concern #5: It introduces matters that are repeated throughout the text but not dealt with in-depth such as the question of ethics in the use of these technologies.
Author response: Thank you for bringing this to our attention. To improve our research's coherence and focus, we have made the decision to remove the discussion on ethics. Instead, we have introduced a new section, "Artificial Intelligence in Architectural Education", which is more pertinent to the overall theme of our paper and reinforces the clarity of our research structure. (Can be found in lines 139 to 177)
Best regards,
Zhihui Zhang
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Dear Editor
My opinions and suggestions for the article titled "Visualizing Gaudí through AI: A Comparative Perception Study of AI-Generated and Real Architectural Designs", which I refereed to, are as follows.
Lines 3-5 Who was it measured by? What is the knowledge of the participants in this study about architecture? The answer to the question is unclear.
Line 9- The explanation of the mentioned study on its relationship with architectural education should be explained in one sentence. How can this study relate to architectural education?
Nowhere in the study has there been a clear methodological inference about how it can be added to architectural education as a method? How can a study that is not carried out in education be integrated into education? There is no information that the participants participating in the study consisted of students. For this reason, "AI in architectural education" stands at an unrelated point throughout the article.
Lines 18-19 The reference in this sentence should be checked. There is a statement that the completion of "AI's transformative" and "Sagrada Familia" is done using AI design tools. Or should it just be "computational design tools"?
Lines 23-25 The cited references do not provide references for architectural design and AI integration. In architectural design, AI studies studied as an intersection in the history of architecture should be researched and the literature part of the introduction should be updated and supported accordingly.
Lines 28-32 The sentence mentioned in 28-29 should be supported with correct references. The references in 31-32 are not working at the intersection of AI and architectural education. Negroponte may be the first example, but there must be another reference from current studies. Reference 7 does not support this narrative. It is recommended to conduct research on architectural education and studies conducted at the interface of AI, which are up-to-date in the literature.
Figure 1 should come right after the text in which it is mentioned.
Lines 87-90 Are networks of GANs a method used to study? If the literature is mentioned, the works produced with GANs networks in architecture should be referenced.
Lines 93-96 References 24,25 are not references to support this narrative. What are the mentioned "AI tools"? "Which networks, algorithms are used?"
Lines 96-98 It is not understood what is meant in the lines of 96-98. It disrupts the flow. Where did we come from environmental data? Why 'environmentally -friendly'?
Lines 107-109 Does the reference given to support this sentence? What is its relationship to AI? The Boden reference needs to be reviewed throughout the text. Not everything computer generated is done using AI tools. Which AI tools? Evolutionary algorithms or Deep Learning?
Figure 3 Which conditioned?
Lines 159-160 Which algorithm was used should be explained more clearly.
Lines 175 What is LoRa? Needs explanation.
Lines 197 which participants?
Lines 220-221 Who are the participants? Background knowledge? Architecture student's experience?
Lines 410-411 educational tools? In what ways can it be used in architecture? What can it contribute to students? How is it envisaged that it can be used for education without being tried in education before?
Lines 420-421 Only its use in architectural design can be written.
Lines 425-426 Which learning style does architectural education contribute to be used in architectural education? These need to be measured and explained. However, the visuals produced in the study were not experienced by the students. Using these expressions may not be correct.
Author Response
Dear Reviewer,
I am profoundly grateful for the time, effort, and expertise that you have dedicated to reviewing our manuscript. Your thoughtful and constructive feedback has significantly enhanced the clarity and rigor of our work, and we feel incredibly fortunate to have received such comprehensive and insightful comments.
We have carefully considered each of your concerns and have addressed them as comprehensively as possible. To assist you in navigating our responses, we have categorised our answers according to the various topics you raised.
This manuscript contains both revised and deleted content and is not yet in its final layout format. Therefore, please forgive any issues regarding alignment of text and images inconsistencies. We will certainly ensure everything is perfectly organized in the final version.
Concern #1: Concerns about the Participants and Measurement Methods:（"Lines 3-5 Who was it measured by? What is the knowledge of the participants in this study about architecture? The answer to the question is unclear." and "Who are the participants? Background knowledge? Architecture student's experience?"）
Author response: We appreciate your feedback and have added quality mechanisms to ensure the validity of our outcomes. Concerning the information on the number of participants and the distribution across different age groups and genders, we have supplemented the necessary data in the manuscript. We concur that the absence of students from the architecture department in our research is regrettable. However, we do not see a necessity to differentiate between males and females, as it is not the aim of our research. Our study primarily explores the possibilities of AI in architectural design and the acceptance level of AI-generated designs by individuals, which is not influenced by gender. (Can be found in lines 238 to 256, lines 457 to 463)
Concern #2: Connection to Architectural Education: （"Line 9- The explanation of the mentioned study on its relationship with architectural education should be explained in one sentence. How can this study relate to architectural education?" and "Lines 410-411 educational tools? In what ways can it be used in architecture? What can it contribute to students? How is it envisaged that it can be used for education without being tried in education before?"）
Author response: Thank you for bringing this to our attention. To improve our research's coherence and focus, we have made the decision to remove the discussion on ethics. Instead, we have introduced a new section, "Artificial Intelligence in Architectural Education", which is more pertinent to the overall theme of our paper and reinforces the clarity of our research structure.
lines 139 to 177，lines 445 to 466
Concern #3: Issues with References and Citations: （"Lines 18-19 The reference in this sentence should be checked. There is a statement that the completion of "AI's transformative" and "Sagrada Familia" is done using AI design tools. Or should it just be "computational design tools"?" and "Lines 28-32 The sentence mentioned in 28-29 should be supported with correct references."）
Author response: Thank you for bringing this to our attention. We have reviewed and updated the references cited in the article to accurately reflect the current status of AI in architectural design.
Concern #4: Queries about AI Tools and Algorithms: （"Lines 93-96 References 24,25 are not references to support this narrative. What are the mentioned "AI tools"? "Which networks, algorithms are used?" and "Lines 159-160 Which algorithm was used should be explained more clearly."）
Author response: Regarding your queries about AI tools and algorithms, we have rewritten the section "Application of AI in the Field of Architecture and Art" (Lines 90-138) and revised the section on the "Creation of AI-Generated Images". We also adjusted Figure 3, to provide a more exhaustive explanation of the networks and algorithms used. You can see these modifications in lines 210-228 of the revised manuscript.
Concern #5: Other Clarification Issues: (“Figure 3 Which conditioned?" and "Lines 175 What is LoRa? Needs explanation.")
Author response: For your other questions that required further explanation or clarification, we have also made the necessary adjustments. We have provided more condition information in Figure 3 to address your question. Moreover, we have explained what LoRA is in the revised manuscript. You can find these modifications in lines 210-228 of the manuscript.
Once again, I extend my sincere thanks for your invaluable critique, which has undoubtedly elevated the quality of our work.
Best wishes
Zhihui Zhang
Author Response File: Author Response.pdf
Reviewer 3 Report
Dear Author/Authors
The article is of interest for architects and bordering disciplines and needed in the field of architecture in all fields and off course in design, education and preservation. The content is well articulated and structured, detailed in analyses and argumentation. Part 3 and 4 are extremely valuable presented clearly and visually appealing. The intro part need a paragraph explaining the overall structure of the article. Before the last paragraph on results.
I have a hold back situation concerning the consistency of the article when it comes to the three chosen specific fields / design, education and preservation. It is very complicated and some times confusing to focus on the research when the investigation is being shifted from education to design and to preservation. I can not see why the author introduces all three fields prior to discussion and conclusions.
It might be wise to focus on architecture in general, giving motives from education and preservation in the intro part (preservation is missing), but also focusing on those specific fields only in discussion and in conclusion.
If introduced from the very start (literature review) of the research, it is important to add a new paragraph 2.1 only on arch. design and AI, then the one already there on Ai and Gaudi with a strong reason why Gaudi and AI, and at the end 2.5 on Ai and arch. preservation, and to expand the whole research focusing on all tree areas consistently in methods and results. It might be a good idea to add background subtitle and transfer 2.1 there and to leave out Gaudi from literature review, as this decision on Gaudi did not come from literature. Then there should be a paragraph where explanation is needed to support that education and preservation are equally important for the topic research as design other fields in general and why the research focuses on these three. Might be a possibility to make a syntheses at the end of literature review and to work with architecture in general from then on. In that way discussion and conclusion part is well structured pointing education and preservation as important points but not specific fields with favorable impact of AI.
This is more a recommendation to think about and decide weather you would consider this structural change as to widen the impact the article can have on the discipline of architecture in general.
Author Response
Dear Reviewer, 3,
Firstly, I want to express my gratitude for your feedback and attention to my research. I have undertaken substantial revisions incorporating both your suggestions and those from other reviewers. Specifically, I have entirely rewritten the literature review section (the revised parts are marked in red) following these themes:
1，AI as an innovative design tool for generating alternative solutions.
2，The potential of AI in architectural education.
3，Assessment of Aesthetic Value in AI-Generated Architectural Design
Comment 1: "The intro part needs a paragraph explaining the overall structure of the article, before the last paragraph on results."
Author Response: We agree with your suggestion and have added a paragraph at the end of the introduction that explains the overall structure of the article. This will help the reader navigate the article better and understand the flow of the arguments. (see Introduction, lines 34)
Comment 2: "I have a hold-back situation concerning the consistency of the article when it comes to the three chosen specific fields / design, education, and preservation. It is very complicated and sometimes confusing to focus on the research when the investigation is being shifted from education to design and to preservation. I cannot see why the author introduces all three fields prior to discussion and conclusions."
Author Response: We understand your concern about the focus of the article. To improve the clarity and consistency of our paper, we have revised the structure and rewrote the parts involving design and education while eliminating the field of preservation. We believe this division will reduce confusion for the readers. (see page 2-3)
Comment 3: "If introduced from the very start (literature review) of the research, it is important to add a new paragraph 2.1 only on architectural design and AI, then the one already there on AI and Gaudí with a strong reason why Gaudí and AI, and at the end 2.5 on AI and architectural preservation, and to expand the whole research focusing on all three areas consistently in methods and results."
Author Response: We appreciate your suggestion and have restructured our literature review as per your advice. We have also removed the part on Gaudí. (see page 2-3)
Comment 4: "This is more a recommendation to think about and decide whether you would consider this structural change as to widen the impact the article can have on the discipline of architecture in general."
Author Response: Your suggestion is indeed valuable. We have made amendments to the article to widen its impact on the discipline of architecture and to ensure its relevance to design among other fields. We believe this widened scope will make our research more useful to a broader readership. (see page16 line 469)
Thank you once again for your thoughtful and helpful comments. We hope our revisions address your concerns and improve the clarity and breadth of our research.
Best regards,
Zhihui Zhang
Round 2
Reviewer 1 Report
The article has introduced some of the considerations that were raised in the previous review, but some major ones have not been fixed. Further work is therefore recommended.
In the previous review, it was recommended to address the problem of risk in the quality of architectural design resulting from an imbalance towards the visual perceptual system. But the authors have merely referred to the issue without reflecting on how it might have affected this study. And the introduction to the multi-sensoriality of architecture hasn't been dealt with relating to this issue.
The question of the definition of the participants has been satisfactorily answered with respect to gender, although it would be interesting to place statistical information on the student's areas of study since it is not a trivial question of how visual perception has been trained depending on this element and therefore to obtain a trustful answer to the question posed about the potential use of artificial intelligence in architectural design.
It was also recommended that a theoretical framework section be included in the article. This section is of vital importance as the analysis of the data is based on a neuroscience model that has not been framed with other models that could have been used as a basis. This is not a minor omission from a scientific paper.
Regarding previous criticism about methodology, the main issues haven’t been fixed. This is a major weakness of the presented study and leads to doubts that the conclusions make a real contribution to knowledge.
Other issues that deserve attention:
Some references to the complexity of architectural design are not very up-to-date, and it would be advisable to include more up-to-date references (see 25).
Some statements are opinionated:
1.- “As a response to the immense data produced about the built environment and human activities, architecture requires substantial support from computational technologies and AI". The use of AI is not a response to that situation as it has always happened. Another position is that technology can help to find solutions from a more objective mode, leaving behind the subjective variable that until now has been submerged in architectural design. Although this is what the authors state in the following paragraph. Therefore, it is recommended to rewrite this part and base it on an existing study.
2.- "...the subsequent part of this review will primarily focus on AI in Architectural Representation, where the transformative potential of AI is most prominent." As we are in the early stages, there is no perspective to state anything yet. It would be more honest to say that this is the area of interest of this study.
3.- "Architecture, although a multisensory experience incorporating elements such as touch, sound, and even smell, primarily captivates through visual expression." In the seventeenth and eighteenth centuries, the Enlightenment period marked a society which prioritised the senses of sight and hearing over the senses of smell, touch and taste ( See Henshaw, (2014). Urban smellscapes: Understanding and designing city smell environments. Routledge.
It is also advisable to consult these other references:
Bille, M., & Sørensen, T. F. (2018). Atmospheric architecture: Elements, processes and practices. In D. Howes (Ed.), Senses and sensation: Critical and primary sources. Bloomsbury.
Sennett, R. (1996). Flesh and Stone: The Body and the City in Western Civilization. WW Norton & Company.
The abstract should include a brief contextualisation and a justification of the problem.
Author Response
Dear Reviewer, 1,
Firstly, I want to express my gratitude for your feedback and attention to my research. I have undertaken substantial revisions incorporating both your suggestions and those from other reviewers. Specifically, I have entirely rewritten the literature review section (the revised parts are marked in red) following these themes:
1，AI as an innovative design tool for generating alternative solutions.
2，The potential of AI in architectural education.
3，Assessment of Aesthetic Value in AI-Generated Architectural Design
Concern #1: "In the previous review, it was recommended to address the problem of risk in the quality of architectural design resulting from an imbalance towards the visual perceptual system."
Author response: We have thoroughly discussed the issue of the visual perceptual system at lines 141 on page 4 and line 428 on page 16, considering how multisensoriality impacts our research. We have included this issue in the research limitations and plan to give it a more comprehensive consideration in future research.
Concern #2: "The question of the definition of the participants has been satisfactorily answered with respect to gender, although it would be interesting to place statistical information on the student's areas of study."
Author response: We recognize the importance of this issue and have added statistical information on the students' fields of study at line 224 on page 7, though we admit the lack of other statistical information due to a design oversight in the experiment.
Concern #3: "It was also recommended that a theoretical framework section be included in the article."
Author response: We appreciate this suggestion. We have now added a theoretical framework section at line 263 on page 8, providing a more comprehensive introduction to the neuroscience model of data analysis.
Concern #4: "Regarding previous criticism about methodology, the main issues haven’t been fixed."
Author response: We acknowledge the significance of this issue and have thoroughly revised the methodology section at line 263 on page 8, ensuring the reliability of the results.
Concern #5: "Some references to the complexity of architectural design are not very up-to-date, and it would be advisable to include more up-to-date references (see 25)."
Author response: We have updated the relevant references.
Concern #6: "Some statements are opinionated."
Author response: We appreciate you pointing out this issue. We have rewritten the literature review section to ensure that our viewpoints are based on existing research, being more neutral and explanatory where possible, on pages 2 to 3.
Concern #7: "The abstract should include a brief contextualisation and a justification of the problem."
Author response: We have revised the abstract on page 1 to include a brief contextualization and justification of the problem.
We hope these responses and modifications meet your expectations. We greatly appreciate your feedback as it has been immensely helpful for our research. We warmly welcome any other issues or suggestions if there are any.
Thank you once again for your time and professional advice.
Best regards,
Zhihui Zhang
Reviewer 2 Report
Dear Editor
I appreciate the edits made by the authors in line with my suggestions. However, since the change is not structural, these touches did not make the article reach a scientific level.
Author Response
Dear Reviewer 2,
Thank you for your time in revisiting our manuscript and for your feedback. We understand that you feel the revisions did not substantially improve the scientific quality of the manuscript.
We have made earnest efforts to incorporate your valuable feedback, and have tried to improve the scientific rigour and clarity of the manuscript. However, we understand that it may still not meet your expectations.
We believe that our study provides a unique perspective on the application of artificial intelligence in the field of architecture. The analysis we present, though specific in its focus on design, education, and aesthetic evaluation, offers important insights that can contribute to the broader architectural discourse.
We respect your expertise and judgement, and we kindly ask if you could provide additional specific points or examples where you think we can further improve the manuscript. Your constructive criticism would be of great help to us in further enhancing the scientific quality of our research.
Thank you once again for your time and effort. We look forward to your continued guidance.
Best regards,
Zhihui Zhang
