CASE STUDY ON SOFTWARE ENGINEERING

AI-Driven Architectural Designs

NAME: KARANAM SAI UDAYA KUMAR

REG.NO: 2021BCSE07AED366

SECTION: ML-C

1. Abstract:
The concept of AI is used to intro the idea of personalized learning, adaptive systems, and efficient applications in an
education platform for very sleek applications which could dethrone the existing. The education applications based on
AI adopt machine learning algorithms, predictive analytics, and natural language processing for analyzing students'
performance data, learning behavior, and preferences in real-time for the delivery of just-in-time content. The SD
model optimizes learning experiences through engagement, understanding, and information retention. The use of
game-based learning brought more interactive tools into this kind of learning platform, such as virtual tutors, chatbots,
simulations, and augmented realities.

AI systems facilitate real-time diagnosis and prediction, thus enabling teachers to identify learning challenges and
intervene accordingly. The benefit of AI systems is their adaptability, promoting inclusiveness for learners from
different backgrounds. This includes those with special needs. Cloud-based platforms allow scaling and access to
different technologies that can support collaborative learning environments without the restrictions of geographical
and socio-economic boundaries.

The challenges facing the implementation of AI in education include violations of data privacy, ethical dilemmas, and
inequitable access. Trust and fairness in these AI-based systems can be assured only by a robust ethical structure and
compliance with data protection regulations. In addition, the integration of AI with other fast-evolving technologies,
such as VR, AR, and Digital twins, will offer great opportunities for enhancing educational outcomes through
immersive and contextualized learning scenarios.

This study investigates AI's transformational forces within educational practices, facilitating innovation address the
needs of diverse learners, and improve efficiency. It is concerned with finding a balance between technology and
ethics.

2. Keywords: AI in education, Personalized learning, Adaptive learning systems, Predictive analytics, Digital
transformation.
3. Introduction:

Fake news contains misleading information that could be checked. This maintenance lies about a particular statistic
in a country or the exaggerated cost of specific services for a country. This may cause unrest in some countries like in
the Arabic Spring. Organizations like the House of Commons and the Crosscheck Project are trying to deal with the
issues by making authors accountable. However, their scope is limited because they rely on human manual detection.
In a world where articles are being removed or published every minute, this can't be accountable or feasible manually.
A solution could be designing a system to provide credible automated indexing and scoring for credibility for different
publishers in the news context.

These proper purposes are a methodology to develop a model that will detect if an article is authentic or fake Based
on its words, phrases, source, and title, using supervised machine learning algorithms. After feature selection methods
are tested, the best-fit feature will be selected to obtain the highest precision from all the confusion matrix results. We
proposed to create a model with different classification algorithms split with the unseen test data that is evaluated, and
the results are plotted which will create a model that detects and marks fake articles that could be integrated with any
further use system..

Fake news detection aims to curb the spread of rumors, especially through social media and messaging platforms.
Such rumors develop into acts of mob lynching. These events have been successfully prompting us to engage further
in this project. Such developments lead to various news or articles on mob lynching, resulting in murder. As a means
of controlling and containing such situations, Fake news detection works to identify and stop such outlandish news as
mob lynching so as to protect society from such an act of aggression.

In today's world, anyone can put out content on the internet. Regrettably, counterfeit news spreads like wildfire across
the web, especially via social media. People get fooled into circulating such disinformation without thinking twice.
Such actions amount to an affront to society when some rumors or vague reports create negativity in society or among
a specific category of people. Contactless systems operate on the same level as information technology. This means it
becomes difficult to control. Societies should take precautions to stop such activities. It broken down into two phases.
Mascommunication is, however, huge and is intended to implement and break the bounds. In such situations, there are
many sites that spread falsehoods. They are deliberately misleading us under the guise of being the true news.
4. Literature survey:

False news detection has thus become one of the most vital areas of research, owing to plethora of misinformation on
digital platforms and their effects on public trust, societal stability, and democracy. These challenges now find a
powerful weapon with the advent of machine learning and natural language processing methods, providing a means
for automated, scalable, and accurate classification of news content. Characterized by fabricating or misleading
information, fake news presents unique problems constantly evolving and threatening to sway public opinion most
often through emotional and sensational language. The research centers on the provision of a robust system dealing
with the problem, relying on integrating high-end algorithms, datasets, and features engineering processes.

The development of an effective fake news detection system relies on the availability of varied and high-quality
datasets. Of these datasets, the Fake News Challenge dataset provides an interface for pairing headlines with articles
for the determination of agreement or disagreement, and the LIAR dataset contains over 12,000 manually labeled
claims by PolitiFact, with six truthfulness levels, including "true," "half-true," and "false." The FakeNewsNet dataset
takes a multimodal approach by integrating news content with social context features like user engagement and sharing
patterns; tempoBuzzFeed News dataset focuses on news articles shared on Facebook to showcase how social media
outlets facilitate misinformation dissemination. Such datasets provide a body for machine learning models' training
and evaluation, but they are constantly challenged by obstacles like class imbalance or linguistic bias.

Feature engineering, which employs schemes for the extraction of meaningful patterns from speech, establishes a
strong base for detective work. The linguistic features include terms of vocabulary such as Frequency-Inverse
Document Frequency (TF-IDF), n-grams, and Bag-of-Words, which help analyze lexical and syntactic structures. It
can provide logico-semantic features through model constructs like Word2Vec, GloVe, FastText, etc., which are surely
instrumental while providing more accurate detections. Language models like BERT and RoBERTa represent modern
state-of-the-art in NLP, where connecting deeply complex textual undertones is addressed with much more claims of
success making the right classification. Use of linguistic features such as hyperbolic language, tenor, and stylistic
schemata provides a strong chance towards the differentiation of real and fake news.

A significant amount of machine learning models have been applied purposely towards fake news detection.
Traditional algorithms such as Logistic Regression, Naïve Bayes, and Support Vector Machines (SVM) have been
found successful for binary classification-type problems-always combined with well-engineered features. Ensemble
methods such as Random Forest, Gradient Boosted Decision Trees (GBDT), contribute to making them more robust
and more accurate due to combining the results of several models. Add to this, meanwhile, Long Short-Term Memory
(LSTM) networks and Convolutional Neural Networks (CNNs), with deep learning approaches trained, focus on
detecting temporal and spatial patterns in the text while transformer approaches such as BERT and RoBERTa
outperform terribly due to frameworks upon contextual dodges and connections.
Multimodal detection techniques leverage textual, visual, and social data for improved fake news detection. Text-and-
image fusion models express inconsistencies between news articles and the corresponding images, which helps to
identify mismatch contents as signals for deception. Social features, including user interactions, sharing patterns, and
comments, also provide insights revealing the way that misinformation spreads across social media platforms. Graph-
based approaches further model the relationships between users and the news articles, revealing clusters of fake news
dissemination across certain groups with influential sources.

But fake news detection, while showing greater completion, still has various complications. Data imbalance will
reasonably continue to plague us, releasing real news to its fake counterparts in a nonbalanced manner and producing
biased model predictions in the process. Techniques including but not limited to oversampling, undersampling, and
synthetic minority oversampling provide possible solutions to the problem. Even more recently, with the intensity and
numerosity of changes to the information published, determining fake news becomes a challenge as the
misinformation evolves so rapidly that the training datasets must always be updated. More importantly, effete
sociopolitical challenges like data privacy, algorithmic bias, and opacity keep scaring users away from trusting these
fake news detection systems and deter most of them from adoption as a technology.

To overcome these challenges, researchers emphasize the importance of creating diverse and representative datasets,
developing robust and adaptive model architectures, and establishing ethical frameworks that prioritize fairness and
accountability. Emerging areas of research, such as multilingual fake news detection, real-time systems, and
blockchain integration for traceability, hold promise for advancing the field. By fostering interdisciplinary
collaboration and leveraging innovative technologies, machine learning-based fake news detection systems can
enhance the reliability of digital information ecosystems and support a more informed and resilient society.

Fog computing is emerging as a key platform to support cellular packages via providing computing, communique, and
garage, specifically in 5G networks. It allows acquire high bandwidth, intelligence, low latency, and energy efficiency
for dealing with huge numbers of gadgets and multimedia offerings. As the point of interest shifts from the Internet of
Things (IoT) to the Internet of Everything (IoE), which connects people, tactics, records, and things, fog radio get
admission to networks (F-RANs) face challenges inclusive of numerous IoE devices, on-demand extremelyexcessive-
definition videos, and occasional-latency desires. This article offers a visible IoE (V-IoE) architecture for F-RANs,
reading key challenges and featuring a crowd V-IoE architecture. Experimental results display that this structure
performs better, requiring much less bandwidth, strength, and latency. The article concludes with capacity future
directions [13].
5.Methodology

5.1 Identify the needs of customers and users through field visits/surveys

Because of the multi-dimensional nature of fake news, recognizing the category of news is not so easy. A practical
technique must contain a few perspectives to precisely handle the issue. This is the reason the proposed strategy is a
mix of Naïve Bayes classifier, Support Vector Machines, and semantic investigation. The proposed strategy is
completely made out of Artificial Intelligence draws near, which is basic to precisely order between the genuine or
the fake, rather than utilizing calculations that can't mirror subjective capacities. The three-section strategy is a blend
between Machine Learning calculations that subdivide into managed learning procedures, and characteristic
language preparing techniques.

5.2 Translate the needs of customers into products.

News aggregator sites enable clients to see news and updates from different sources in one advantageous area. They
get the information, arrange it in labels/classes, and show it organized appropriately for simpler utilization. There are
various top sites like Google News, Feedly, news360, etc. which provide semi-structured news data. The RSS
Aggregator plugins are available to make things easy. Aggregator builds the quality and accuracy of/ the news. Data
gathering is the main purpose of any news aggregator. A basic methodology is consistently checking RSS Feeds;
Article extraction from various news sites and gathering information. Regularly applied systems to discover related
articles are watchword-based methodologies. At the point when all procedure is done, they show important or
ongoing news on the page.

5.3 Specification of the product

A new authenticator follows some steps to check whether the news is true or false. It will compare news that is given
by our side with different websites and various news sources if that news is found on any news website then it shows
the given news is true, else it shows there has been no such news in the last few days. This can help us from fake
news. These days‟ fake news spreads very fast because of social media and the internet. So, a news authenticator
helps us to detect whether the given news is fake or real.

The private insurance sector is swiftly developing, with insurance merchandise now overlaying high-price belongings
like cars, jewelry, fitness, and homes. To maintain up, coverage companies are adopting superior operations, processes,
and mathematical fashions to maximize profits and enhance customer service. Traditional techniques, relying closely
on human involvement, are sluggish and frequently faulty. This paper offers a stable, automatic coverage system that
reduces human interaction, enhances security, detects risky customers and fraudulent claims, and minimizes losses.
The framework is carried out the use of system getting to know modules and Hyperledger Fabric Composer, combining
AI and blockchain technologies [16].
5.4 Designing the solution for required customer needs.

News suggestions suggest recent news and news related to the news that the user has
given for authentication. If the news is fake, then this news suggestion gives the related
newson that topic. The news suggestion suggests the news based on keywords that you
give in your news that you wish to authenticate.

5.5 Web Application (Tech stack)

1 Frontend

a HTML

b Tailwind CSS
c Tail blocks

d Un plash

2 Backend
1.1. Flask (Python)
1.2. Vs Code (IDE)
1.3. Git & GitHub (Version Control)

3 Machine Learning

a Pandas

b NumPy

c Confusion matrix

d Pickle

e Jupiter Notebook (IDE)

f. Deployment
6 Implementation

Architecture for fake news detection

Output : Fake
“Fake Output" in fake news detection using machine learning consists of a clear label indicating the article's
classification as fake, accompanied by a confidence score and an explanation of the classification. Visual indicators,
actionable insights, and feedback mechanisms are provided to help users understand and respond to the
classification effectively

Output: Real

A "true output" in fake news detection using machine learning confirms the news article's authenticity with a clear
label and a confidence score. It includes an explanation of the classification, visual indicators, actionable insights, and
a feedback mechanism to enhance user understanding and trust in the model's predictions
7 CONCLUSIONS

It is crucial to assess the accuracy of news available on the internet. This paper discusses the elements involved in
identifying fake news. It's important to note that not all fake news spreads through social media. Currently, the
proposed method utilizes Naïve Bayes classifiers, SVM, and NLP for testing. In the future, the resulting algorithm
may yield improved outcomes through hybrid approaches aimed at the same goal. The system mentioned detects fake
news based on the applied models and also offers suggested articles on the topic, which can be very beneficial for
users. Moving forward, the efficiency and accuracy of the prototype can be enhanced, along with improvements to the
user interface of the proposed model. With the rise in internet usage, spreading fake news has become easier. A large
number of people are constantly connected to the internet and social media platforms, where there are no restrictions
on posting news. This lack of regulation allows some individuals to exploit these platforms to disseminate false
information about people or organizations, potentially damaging reputations or impacting businesses. Fake news can
also sway public opinion regarding political parties. Therefore, there is a pressing need for effective detection methods.
Machine learning classifiers serve various purposes and can also be employed to identify fake news. Initially, these
classifiers are trained using a dataset known as the training dataset, enabling them to automatically recognize fake
news thereafter. This systematic literature review focuses on supervised machine learning classifiers that require
labeled data for training. However, labeled data is not readily available, which poses a challenge for training classifiers
to detect fake news. Future research could explore the application of unsupervised machine learning classifiers for this
purpose.
8 REFERENCES

1. Abdullah-All-Tanvir, Mahir, E. M., Akhter S., & Huq, M. R. (2019). Detecting Fake News using Machine
Learning and Deep Learning Algorithms. 7th International Conference on Smart Computing &
Communications (ICSCC), Sarawak, Malaysia, Malaysia, 2019, pp.1-5,

https://doi.org/10.1109/ICSCC.2019.8843612 Ahmed, H., Traore, I., & Saad, S. (2017).

2. Detection of online fake news using n-gram analysis and machine learning techniques. Proceedings of the
International Conference on Intelligent, Secure, and Dependable Systems in Distributed and Cloud
Environments, 127–138, Springer, Vancouver, Canada, 2017.

https://doi.org/10.1007/978-3-319-69155-8_9 Ahmed, H., Traoré, I., & Saad, S. (2018).

3. Detecting opinion spam and fake news using text classification. Secure. Priv., 1(1), 1-

15. https://doi.org/10.1002/spy2.9 Al Asaad, B., & Erascu, M. (2018). A Tool for Fake News Detection. 2018
20th International Symposium on Symbolic and Numeric Algorithms for Scientific Computing (SYNASC),
Timisoara, Romania, 2018, pp.379-

386. https://doi.org/10.1109/SYNASC.2018.00064 Aphiwongsophon, S., & Chongstitvatana, P. (2018).

4. Detecting Fake News with Machine Learning Method. 2018 15th International Conference on
Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology
(ECTI-CON), 528-531. https://doi.org/10.1109/ECTICon.2018.8620051 Della Vedova, M. L.,
Tacchini, E., Moret, S., Ballarin, G., DiPierro, M., & de Alfaro, L. (2018).

5. Automatic online fake news detection combining content and social signals. FRUCT'22: Proceedings of
the 22nd Conference of Open Innovations Association FRUCT. Pages 272–279.

https://dl.acm.org/doi/10.5555/3266365.3266403 Dewey, C. (2016).

9 Data Feature Representation and Dataset Construction
In the improvement of structural wise layout, how facts capabilities are represented plays a pivotal function. These
records functions embody quite a few characteristics, such as geometric and mechanical aspects of the structural
layout. The representation of those features must adhere to the constraints of deep gaining knowledge of algorithms,
which could manner full-size quantities of statistics but are restrained by way of the first-class and structure of the
functions provided. In constructing structural design, the primary features consist of topological, pattern, and size
functions. Topological functions are involved with the spatial format of structural additives and the connection
relationships between those additives. Pattern capabilities capture habitual styles or configurations in design facts,
that may assist in automating the era of similar designs and identifying green design solutions. Constructing a dataset
with those capabilities requires careful consideration of ways the facts is prepared, formatted, and processed to make
sure compatibility with deep studying models.

ii. Generative AI Algorithms for Structural Design

Generative AI has emerged as a effective tool for structural format, supplying the potential to investigate from high-
quality datasets of past design projects, mechanical necessities, and empirical tips. Different kinds of building
structures, together with residential, industrial, and industrial, have their very own layout guidelines and suggestions,
requiring the development of specialized generative AI algorithms tailored to these precise competencies. Generative
AI leverages tool studying strategies to apprehend the patterns and ideas that underlie a success structural design. By
studying from huge datasets of historical designs, the ones AI models can generate modern-day new designs that meet
hooked up requirements for safety, capability, and overall performance. These algorithms are especially beneficial in
eventualities wherein conventional methods of layout may be time-consuming or inefficient, as generative AI can
quick produce and compare a sizeable sort of format options.

iii.

Evaluation Methods for Generative AI Design

This evaluation facilitates to ensure that the generated designs meet the specified safety standards, aesthetic
expectancies, and sensible dreams. The assessment method typically includes more than one stages, inclusive of
assessing the design at some point of both the training and checking out degrees of the AI model. During the schooling
phase, the assessment makes a specialty of the AI’s capability to observe from the provided records and generate
accurate predictions or layout solutions. The loss characteristic need to be differentiable, permitting the version to
adjust and decorate at some stage in the schooling system. In the checking out segment, the designs are subjected to
several actual-global constraints and situations, along with load-bearing capacity, cloth charges, and regulatory
requirements.

iv. Integration of Intelligent Generation and Optimization

Generative AI provides a strong approach for speedy producing new layout options based totally on present
information and discovered guidelines. While the AI can generate a big variety of layout answers, these answers won't
always definitely fulfill the accuracy or precision required via format specifications, especially whilst dealing with
mainly targeted or localized structural requirements. Optimization techniques can firstrate-track AI-generated designs
to make certain they meet all important requirements, collectively with structural integrity, rate-effectiveness, and
compliance with constructing codes. By combining the strengths of both generative AI and optimization, the format
way will become more efficient, ensuring great consequences that are every innovative and accurate.

The aggregate of generative AI and optimization strategies allows for a extra holistic technique to design. Generative
AI brings creativity and speed to the table, quick generating a vast range of solutions, even as optimization refines
those designs to make them realistic, accurate, and compliant with all vital standards. This synergy no longer best
speeds up the design system but additionally ensures that the results aren't just innovative and visually attractive but
also structurally sound, fee-effective, and legally compliant.

v. Development of Intelligent Structural Design Systems

The improvement of shrewd structural layout structures that integrate generative AI is revolutionizing the manner
houses are designed. These systems permit clients to go into architectural and layout parameters, which includes the
layout of shear walls, beams, and columns, further to precise format situations like loadbearing ability and
environmental elements. The generative AI then approaches those inputs and generates corresponding structural
schemes. This cloud-based totally layout system no longer best automates tons of the format method however also
lets in for real-time collaboration among stakeholders, consisting of architects, engineers, and contractors. It also
enables non-stop development by means of permitting designers to regulate and refine their inputs as desired. One of
the important benefits of those structures is their capability to automate big quantities of the design method.
Traditionally, structural layout worried meticulous manual calculations and numerous iterations, which may be time-
eating and at risk of human blunders. With the help of generative AI, this method turns into an awful lot more
streamlined. The AI constantly analyzes the provided parameters and generates a variety of structural options in a
fragment of the time, presenting modern solutions that won't had been taken into consideration thru conventional
layout methods. This acceleration of the layout manner substantially reduces the time-to-market for tasks, main to
more green workflows and faster assignment completion.

Another important issue of these structures is their cloud-based totally nature. Cloud generation allows real-time
collaboration amongst a couple of stakeholders—together with architects, structural engineers, contractors, or even
clients—who can all get right of entry to and contribute to the design system from unique places. This interconnected
surroundings promotes seamless communique and immediate remarks, ensuring that everyone events are aligned all
through the assignment. It fosters higher choice-making, as teams can instantly adjust primarily based on the AI-
generated options or refine the layout primarily based on stakeholder input.
vi. Level of Generative AI Design for Building Structure

As generative AI continues to evolve inside the discipline of structural format, its involvement may be categorized
into hierarchical ranges ranging from L0 to L5. These tiers replicate the developing diploma of automation and
intelligence blanketed into the design method. At L0, the layout manner is completely guide, with engineers relying
absolutely on their knowledge to create building systems. By L2, AI-generated designs end up extra complicated, with
the device able to be optimizing fantastic components of the layout. At L3, AI can autonomously generate entire
format schemes primarily based on input parameters and venture goals, with minimal human intervention. L4
represents the point at which AI completely integrates into the design manner, generating quite optimized answers at
the same time as nevertheless taking into consideration human enter and oversight. This level of AI layout holds the
capability to revolutionize the sphere of structure and engineering via using allowing faster, extra correct, and further
sustainable layout solutions.
The implementation of an AI-based architectural layout gadget entails a consumer-pleasant interface wherein
architects and engineers input layout necessities, guiding the method. Generative AI fashions analyze those inputs the
use of pre-skilled datasets, leveraging extracted capabilities which includes geometry, topology, and mechanical traits
to generate innovative designs. A sturdy database supports the machine by using storing design facts, structural
standards, and training materials, while facts processing ensures that inputs are wiped clean and organized for AI
analysis. The generated designs go through validation to satisfy person-described criteria and structural norms, with
assessment and feedback mechanisms refining each the fashions and outputs. Centralized cloud storage permits secure
facts get right of entry to and collaboration, whilst the very last validated designs, which include 3-d fashions and
layouts, are brought for further use.

Fig-1: Flowchart on ai-driven architectural designs

7. Results:
The implementation of an AI-primarily based architectural layout machine revolutionizes traditional design methods
by means of combining generative AI fashions with sensible records processing. These systems make use of feature
extraction to investigate spatial layouts, structural interconnectivity, and geometric characteristics, making sure that
each layout reflects architectural and engineering standards. Design validation mechanisms look at those outputs
towards structural, aesthetic, and regulatory benchmarks, making sure sensible and reliable consequences.
Additionally, the integration of strong databases and cloud garage systems helps actual-time collaboration amongst
stakeholders, retaining consistency and accessibility all through the design lifecycle.

Fig-2: Testing and Validation of design score.

Fig-3: Prediction Errors (Residuals)

 Fig-4: Actual Sustainability Scores

Fig-5: Feature Importance Plot for bar graph.

Real-time evaluation and remarks mechanisms permit non-prevent refinement of the designs, ensuring that the AI
models examine and enhance iteratively. The machine’s ability to analyze character inputs, optimize designs, and adapt
to converting necessities reduces guide mistakes and extensively speeds up the format technique. The result is a
harmonious blend of innovation and capability—architectural designs which is probably visually compelling,
structurally inexperienced, environmentally sustainable, and perfectly tailored to purchaser goals. This transformative
technique not most effective complements performance but additionally paves the way for smarter, facts-pushed
architectural practices.
8. Conclusion:
The integration of AI into architectural layout is not best a technological development but a essential shift in how we
envision and collect our built environment. AI-based totally genuinely virtually systems allow architects to gadget
large datasets, extract giant insights, and generate current designs that have been previously unthinkable. By combining
generative AI fashions, client-wonderful interfaces, and actual-time comments mechanisms, the ones structures
enhance creativity, streamline strategies, and make sure compliance with technical requirements. The adaptability of
AI allows the creation of sustainable, fee-green, and structurally sound designs tailor-made to numerous desires and
stressful conditions.

AI’s feature extends beyond layout optimization—it transforms architectural workflows via automating repetitive
responsibilities, lowering design errors, and accelerating selection-making. Generative AI allows architects to find out
endless layout iterations inside seconds, factoring in parameters like web page conditions, strength performance,
material utilization, and aesthetics. Additionally, AI lets in architects to simulate how designs will perform underneath
real-international conditions, which includes wind masses, temperature fluctuations, and seismic interest, making sure
resilience and safety whilst adhering to green building requirements.

Despite its transformative capacity, the implementation of AI in architecture gives large challenges. Integration with
Internet of Things (IoT) technology will similarly enhance smart constructing systems, optimizing energy
consumption, tracking environmental situations, and improving occupant consolation in actual time. AI-pushed
solutions will even deal with pressing global demanding situations, which include urbanization, weather alternate, and
useful resource scarcity, by way of selling round economies and reducing production waste through efficient material
utilization.

In the feature, advancements in AI should revolutionize architectural education, equipping designers with gear to
experiment with complex geometries, predict lengthy-time period building overall performance, and explore
sustainable solutions extra successfully. Furthermore, the combination of AI with rising technologies like digital twins,
robotics, and advanced 3D printing will allow the advent of custom designed, prefabricated structures that are each
progressive and aid efficient.

However, while the capability is massive, real-international implementation comes with annoying conditions on the
facet of making sure information accuracy, managing the probabilistic nature of AI outputs, and preserving a balance
amongst automation and human know-how. Despite those disturbing conditions, AI is paving the manner for smarter
cities, more sustainable creation, and collaborative format techniques that merge human intuition with system
precision.

9. References:
[1] Liao, W., Lu, X., Fei, Y., Gu, Y., & Huang, Y. (2024). Generative AI design for building structures.
Automation in Construction, 157, 105187.

[2] Ploennigs, J. and Berger, M., (2023). AI art in architecture. AI in Civil Engineering, 2(1), p.8.

[3] Rafsanjani, H. N., & Nabizadeh, A. H. (2023). Towards human-centered artificial intelligence (AI) in
architecture, engineering, and construction (AEC) industry. Computers in Human Behavior Reports, 100319.

[4] Liu, Y., Lo, S. K., Lu, Q., Zhu, L., Zhao, D., Xu, X., ... & Whittle, J. (2025). Agent design pattern catalogue:
A collection of architectural patterns for foundation model-based agents. Journal of Systems and Software,
220, 112278.

[5] Ojelabi, E. T., & Adekunle, E. O. (2024). AI-Driven Building Architecture: Ethics, Sustainability and
Optimization. Brolly, 5(2), 235-248.

[6] Abbasabadi, N., & Ashayeri, M. (Eds.). (2024). Artificial Intelligence in Performance-Driven Design:
Theories, Methods, and Tools. John Wiley & Sons.

[7] Rane, N., Choudhary, S., & Rane, J. (2023). Leading-edge technologies for architectural design: a
comprehensive review. Available at SSRN 4637891.

[8] Heyn, H. M., Knauss, E., & Pelliccione, P. (2023). A compositional approach to creating architecture
frameworks with an application to distributed AI systems. Journal of Systems and Software, 198, 111604.

[9] Zhou, S., Sun, J., Xu, K., & Wang, G. (2024). AI-driven data processing and decision optimization in IoT
through edge computing and cloud architecture. Journal of AI-Powered Medical Innovations (International
online ISSN 3078-1930), 2(1), 64-92.

[10] Long, L. D. (2023). An AI-driven model for predicting and optimizing energy-efficient building envelopes.
Alexandria Engineering Journal, 79, 480-501.

[11] Caiazzo, B., Murino, T., Petrillo, A., Piccirillo, G., & Santini, S. (2023). An IoT-based and cloud-assisted
AIdriven monitoring platform for smart manufacturing: design architecture and experimental validation.
Journal of Manufacturing Technology Management, 34(4), 507-534.

[12] Aydemir, A. Z., & Jacoby, S. (2022). Architectural design research: Drivers of practice. The Design Journal,
25(4), 657-674.

[13] Ji, W., Liang, B., Wang, Y., Qiu, R., & Yang, Z. (2020). Crowd V-IoE: Visual internet of everything
architecture in AI-driven fog computing. IEEE Wireless Communications, 27(2), 51-57.

[14] Qiao, X., Huang, Y., Dustdar, S., & Chen, J. (2020). 6G vision: An AI-driven decentralized network and
service architecture. IEEE Internet Computing, 24(4), 33-40.

[15] Liang, C., Du, H., Sun, Y., Niyato, D., Kang, J., Zhao, D., & Imran, M. A. (2024). Generative AI-driven
semantic communication networks: Architecture, technologies and applications. IEEE Transactions on
Cognitive Communications and Networking.

[16] Dhieb, N., Ghazzai, H., Besbes, H., & Massoud, Y. (2020). A secure ai-driven architecture for automated
insurance systems: Fraud detection and risk measurement. IEEE Access, 8, 58546-58558.

[17] Spyrou, O., Hurst, W., & Krampe, C. (2025). A Reference Architecture for Virtual Human Integration in the
Metaverse: Enhancing the Galleries, Libraries, Archives, and Museums (GLAM) Sector with AI-Driven
Experiences. Future Internet, 17(1), 36.
[18] Prakash, S., Cheng, A., Yik, J., Tschand, A., Ghosal, R., Uchendu, I., ... & Reddi, V. J. (2025). QuArch: A
Question-Answering Dataset for AI Agents in Computer Architecture. arXiv preprint arXiv:2501.01892.

[19] Zhang, H., Falletta, N. J., Xie, J., Yu, R., Lee, S., Billah, S. M., & Carroll, J. M. (2025, January). Enhancing
the Travel Experience for People with Visual Impairments through Multimodal Interaction: NaviGPT, A Real-
Time AI-Driven Mobile Navigation System. In The 2025 ACM International Conference on Supporting
Group Work (pp. 29-35).

[20] Newton, D. (2019). Generative deep learning in architectural design. Technology| Architecture+ Design, 3(2),
176-189.

[21] Abouelyazid, M., & Xiang, C. (2019). Architectures for AI Integration in Next-Generation Cloud
Infrastructure, Development, Security, and Management. International Journal of Information and
Cybersecurity, 3(1), 1-19.

[22] As, I., Pal, S., & Basu, P. (2018). Artificial intelligence in architecture: Generating conceptual design via deep
learning. International Journal of Architectural Computing, 16(4), 306-327.

[23] Mostafa, F., Tao, L., & Yu, W. (2021). An effective architecture of digital twin system to support human
decision making and AI‐driven autonomy. Concurrency and Computation: Practice and Experience, 33(19),
e6111.