Academic

Module Assessment Type

Year

A21 6CS014 - Complex System Individual Report

Ai in Architecture

Student Id : 2049834
Student Name : Sulav Shrestha
Section : L6CG4
Module Leader : Mr. Rupak Koirala

Submitted on : 2022-01-02

Acknowledgement
I would like to express my special thanks of gratitude to my teacher Mr. Rupak Koirala who gave
me the golden opportunity to do this report , which also helped me in doing a lot of Research and I
came to know about so many new things I am really thankful to him. Secondly I would also like to
thank my parents and friends who helped me a lot in finalizing this project within the limited time
frame. 

Contents
Introduction..........................................................................................................................................1
Aim and Objective...............................................................................................................................1
Literature Review.................................................................................................................................2
Computer-aided design (CAD)........................................................................................................2
 Generative Design............................................................................................................................3
Building Information Modelling (BIM)...........................................................................................4
Artificial intelligence Capability to Design.....................................................................................5
Analysis of Findings.............................................................................................................................6
Conclusion............................................................................................................................................7
Table of Figures
Figure 1 Designing a house in CAD....................................................................................................3
Figure 2 Chair created with generative design.....................................................................................4
Figure 3 BIM information model.........................................................................................................5
Introduction
Artificial Intelligence (AI) has already made inroads into the industry, allowing it to tackle new
challenges. Although its use in architecture is still in its early stages, preliminary results are
promising. This technology is far more than a coincidence; it is undeniably a big step forward,
capable of profoundly altering architectural practice. This exhibition examines how humans interact
with their surroundings. Understanding AI's impact on architecture necessitates first defining it,
then describing what it involves in terms of techniques and paradigms. First AI must be considered
in the context of scientific and architectural history. Rather than a "disruption," the technological
growth surrounding and supporting AI is the result of a long maturation. The profession has been
undergoing a transformation for quite some time. The gradual adoption of technological solutions
has already profoundly altered each stage of its value chain: first, through the use of new
construction techniques, then through the development of appropriate design software, and finally,
through the introduction of statistical computing capabilities, which include, at the forefront, data
science and artificial intelligence. We'd prefer to see a trajectory shift whose acceleration is
anchored in the continuity of a process that has taken architecture to where it is now, rather than a
sudden large change.

Aim and Objective

Architecture has continued to develop over the years but due to the continued increasing demands
and increasing population architecture development is not fast enough. Due to the increasing
population we need to be able to create buildings that hold a lot of people while having a healthy
living place and capable of holding families. With the help of artificial intelligence, the
development of the architecture can be faster and be relevant with the need of people. Artificial
intelligence wont completely replace architects but, enhance their performance. Architects will be
able to solve problems faster, obtain optimal design and be able to come up with newer ideas that
they might not have ever thought about. The pace of development for architecture will be able to
keep up with the pace with the needs of people. What parts of architecture will AI be able to
improve upon.

By including artificial intelligence in architecture, we hope to improve on the existing knowledge

about architecture while having real people design because without human involvement the designs
would lack human feel and lose its uniqueness. Automation is also one of the bigger topics in this
report as their many repetitive tasks during the building process. We also don’t know if it’s a good
idea to let AI design our homes and products or if the human mind is only thing that is capable
creating things that has it own identity.

1
Literature Review
Since its inception, Al has encompassed semantic networks, search and optimization, rule-based
systems, frames, constraints propagation, backtracking, planning, picture and natural language, and
learning. In recent years, "learning" has been associated with Al and referred to by the general
public as "machine learning (ML)”. AI's application to architecture expanded greatly in the 1960s
thanks to the work of scientists such as Ivan Sutherland, who created the first computer-aided
design (CAD) system dubbed "Sketchpad" and Herbert Simon's classic book The Sciences of the
Artificial In the decades that followed, a number of writers employed a variety of AI techniques and
concepts to solve architectural problems. (Naveen K. Muthumanickam, n.d.)

Currently those works are still being optimized. The performance of prospective solutions was
simulated using sophisticated software from a variety of perspectives, including structural and
environmental and multicriteria optimization to create solutions for different projects that fit in
different criteria. Machine learning is also being used an optimizer to by looking at previous
information and increase the speed of gathering data to build upon it. Artificial intelligence will
help improve many technologies that are used in architecture (Naveen K. Muthumanickam, n.d.).
Architects use many tools to do their works. Some the major technology they use are BIM
(Building Information Modelling), Generative Design and CAD(Computer-Aided Design). (Naveen
K. Muthumanickam, n.d.)

Computer-aided design (CAD)

Computer-aided design (CAD) is a method of creating 2D and 3D models of real-world goods

digitally before they are created. You can simply share, evaluate, simulate, and alter designs with
3D CAD, allowing you to create unique and differentiated goods that hit the market quickly (ptc,
n.d.). With the help of CAD architects are able to plan out their designs while perfectly replicating
the end product to check for any defects or if they add extra parts to their designs. To make this
process more effective researchers have been trying to integrate AI and machine learning to CAD.
(TOMIYAMA, 2007)

The objective of intelligent CAD development was to create a CAD system that could design
objects with minimal user inputs and interactions. Design encompassed a wide range of activities,
including analysis, component or material selection, parametric design, optimization, data integrity
management (including geometric constraint management), process planning, and synthesis. Many
knowledge-based systems were designed with a specific use in mind, and thus displayed a range of
"intelligent" design characteristics. These systems were tested in the field, and some of them were
able to achieve what they set out to do: partial design automation, primarily in parametric design
and selection jobs with constraint management. However, many problems were revealed after using
it. First, it wasn’t capable of creating new designs but only recreate or revise old design. Second,
Second, these systems tended to focus on a narrow domain, sometimes even an inappropriately
small domain, requiring just a specific model to represent design processes. Other problems include
not being able to manage and update large knowledge base. But after years of research with many
trials and error a fully applicable intelligent is not possible but with the advancement of internet
technology it might be far off. (TOMIYAMA, 2007)
Figure 1 Designing a house in CAD

Generative Design

Generative design is a design exploration approach that employs artificial intelligence to generate a
wide range of answers and ideas for complicated issues. The next frontier in computer-aided design
is often regarded as generative design. It uses AI to create high-performance and extremely
intelligent design iterations that aid in the resolution of challenging challenges such as decreasing
component weight, optimizing performance, and simplifying designs. Thousands of design choices
can be explored, validated, and compared at the same time using generative design. The engineer or
designer then evaluates these and makes the ultimate decision. (Aparaschivei, 2021)

The key reasons for employing generative design (GD) technology in architecture are to use
computational capabilities to assist human designers and/or to automate aspects of the design
process. Aside from achieving efficiency (many design instances in a short amount of time), cost
reduction (less time and labor), optimization, correctness, consistency, and other objectives, one of
the primary goals is to explore a larger design space and enable design innovation. Generative
design is an iterative design process that involves a program that creates a certain number of outputs
that meet specific criteria and a designer who fine-tunes the feasible region by selecting individual
outputs or altering input values, ranges, and distribution. A review of the literature revealed that the
bulk of extant GD systems are based on one of these GD approaches. This constraint frequently
jeopardizes the system's performance in terms of design research scope. Although there are some
parallels and overlaps, each of these solutions appears to be better suited for specific design goals
than others. This study argues for the necessity for a unified GD framework that can support a wide
range of approaches. The integrated GD framework is intended to be used as a system for
interactive design generation. Unlike most existing GD systems, which are limited to a particular
GD technique and frequently bias the GD process in a specific direction, the proposed integrated
GD framework can give design triggers at various stages of design development and aid in design
exploration across numerous pathways. (VishalSingh, 2012).
Figure 2 Chair created with generative design

Building Information Modelling (BIM)

BIM is a technique for creating and managing information on a construction project throughout its
life cycle. As part of this method, a coordinated digital description of every characteristic of the
created asset is generated using a set of relevant technology. This digital description will most likely
include a combination of information-rich 3D models and associated structured data such as
product, execution, and handover information. (Hamil, 2021). Researchers in architecture,
engineering, and construction have long sought to use information technology, robots, and other
developing technologies to design and build. However, their conceptual understandings of what
could be done, and hence their visionary perspectives of construction's future, ultimately
outstripped the practical, technological, commercial, cultural, and/or organizational limits that had
to be overcome in order for them to be realized. The requirement to differentiate between modeling
geometry in two and three dimensions led to the creation of BIM dimensions. This has been a part
of the modeling progression, from drawing boards to the earliest 2D CAD systems to 3D modeling
software. (Rafael Sacks, 2020)

Automation of design and code-compliance inspections for building construction has been a
research and commercial development aim since Eastman's pioneering 1975 study envisioning
BIM. In the absence of BIM, researchers proposed stand-alone expert systems and, later, systems
that represented structures using CAD drawings. HI-RISE for preliminary structural design of tall
structures, SPEX for sizing structural cross-sections, and EIDOCC for design of reinforced concrete
columns were utilized in the former. Most of these were rule-based systems that attempted to collect
expert knowledge, store it in design software, and then use it to automate or evaluate design. These
methods were unsuitable for commercial usage due to the required to explicitly and completely
input construction designs for each research, the restrictions on knowledge elicitation, and the
computing technology's capability. Because CAD's graphic design representations differ
fundamentally from the semantic, object-oriented representations required for processing rules, the
development of 2D CAD effectively solved the problems. (Rafael Sacks, 2020)

The widespread use of BIM systems in the construction sector is insufficient to enable effective
exploitation of the data stored within them, much alone capitalize on the potential of AI in this
context. The issue is that the data offered by models is insufficient and difficult to retrieve. We
identified two critical research concerns related to the development of foundational information
processing methods for digital building information models that, if resolved, will greatly facilitate
the development of smart BIM and AI tools for design and construction (Rafael Sacks, 2020). They
are:
 Optimal usage of topological rule inferencing and machine learning modules in conjunction
for semantic enrichment (Rafael Sacks, 2020)
 Encoding representations of building information in machine-learning-friendly formats
(Rafael Sacks, 2020)

In the future, a golden thread of knowledge will be generated collectively in parallel with the design
and development of built assets. This will be a record of what was built as well as how the asset is
now functioning. This will eventually be the "big data" that allows decision makers to constantly
improve in order to produce a safer and more sustainable built environment across client estate – or
even national – boundaries. (Hamil, 2021)

Figure 3 BIM information model

Artificial intelligence Capability to Design

A question develops as a result of all of the study. Is it possible for computers to assist designers in
the production of objects, structures, and architecture, or should the designing be left to humans?
Although both computers and design are types of mental activity, there are significant differences
between them. Computers are part of a long human heritage of attempting to externalize thoughts so
that they may be transferred between people as overt knowledge. This inheritance includes natural
scientific objectivity, mathematical detachment, and formal logic facts. It is concerned with
knowledge that can only be communicated symbolically in abstract forms such as words and
numbers. It is now well acknowledged for establishing the groundwork for our rationalizations and
excuses for everything we do. Computers are the most recent technical innovation in human history,
providing environments in which information may be represented independently of individuals and
act autonomously on their behalf or in their stead.

Design, on the other hand, is distinguished by its use of people's intuitive knowledge, which
encompasses all sorts of human experience and necessitates people's holistic involvement beyond
their exercise of formal, overt knowledge. Individual judgment is employed in design to inform us
about what we wish to do. This contrast can be reinforced by stressing that design work necessitates
human sensitivity, perception, and interpretation abilities that are not constrained by standards that
limit the validity of symbolic expressions. Design typically employs analogic ways of expression,
involving the direct translation of people's concerns into physical items or, more broadly, in
decisions affecting our exploitation of our material world. Correct or, more precisely, good designs
are chosen, but they are never tested. (A., 1986)

Can we be sure about any distinction that we draw between these two forms of knowledge? If we
are to assume that there is no difference between a human mind and artificial intelligence that will
also support the fact that the activity of a human mental can copied by an artificial which seems
implausible and way out of our current capabilities. A proper assumption would be that artificial
intelligence can’t copy a human mental but can mimic human like properties that can only perform
simple tasks like revising or improving on the existing design with the help of preexisting
knowledge. (A., 1986)

Analysis of Findings
Earlier, concerns about the development of architecture being too slow for the growing needs of
people was raised. It has been shown that by including AI many parts of architecture processes can
be made easier and done faster. Like, by including intelligent Computer-aided design(CAD) we can
create digital replicas with very low amount of input and interruptions. We have also seen that the
somethings may not be possible with our current technology and researches are still being
conducted on this topic. To make works more efficient architects have implemented generative
design will helps to employ computational capabilities to assist human designers and (or) to
automate portions of the design process. It also discussed that is it really ok for AI to design
everything. It was conclude that AI can only mimic human mental and do task revising and
improving on existing designs and only a human could do good design because good designs are
chosen, but they are never tested. (Rafael Sacks, 2020)
Conclusion
In the end, we can say that have by integrating AI into architecture its development has been
increased by two folds and still newer discoveries are being made. Things like CAD, BIM, and
generative design are still being researched each year and are being improved upon by the years.
But there are still many things that not possible with our current knowledge and are really not sure
about when we will be able solve them. But it is clear that with recent development in artificial
intelligence and internet technology sooner or later will be able to figure the things that still
understood and solved about architecture in AI.
References
A., B., 1986. AI in architectural CAD.. CAD and Robotics in Architecture and Construction.
Aparaschivei, L., 2021. vectornato. [Online]
Available at: https://www.vectornator.io/blog/generative-design
[Accessed 2021].
Fisher, T., 2020. lifewire.com. [Online]
Available at: https://www.lifewire.com/power-supply-unit-2618158
[Accessed 03 04 2020].
Hamil, D. S., 2021. NBS. [Online]
Available at: https://www.thenbs.com/knowledge/what-is-building-information-modelling-bim
[Accessed 2022].
Naveen K. Muthumanickam, J. P. D. S. N. A. M. a. S. G. B., n.d. Case studies of AI in architecture,
s.l.: s.n.
ptc, n.d. ptc. [Online]
Available at: https://www.ptc.com/en/technologies/cad
[Accessed 27 12 2021].
Rafael Sacks, M. G. I. B., 2020. Building Information Modelling. Artificial Intelligence and
Construction Tech, Volume 4.
Silwen, 2019. Turbofuture.com. [Online]
Available at: https://turbofuture.com/computers/What-are-the-basic-functions-of-a-CPU
[Accessed 29 03 2020].
TOMIYAMA, T., 2007. Past 20 years and future 20 years. Artificial Intelligence for Engineering
Design, Analysis and Manufacturing. Intelligent computer-aided design systems, 21(1), pp. 27-29.
TutorialsWeb, 2019. TutorialsWeb.com. [Online]
Available at: https://www.tutorialsweb.com/computers/pc-motherboard-1.htm
[Accessed 29 03 2020].
VishalSingh, N., 2012. Towards an integrated generative design framework. Design Studies, 33(2),
pp. 185-207.
Zandbergen, P., 2019. study.com. [Online]
Available at: https://study.com/academy/lesson/central-processing-unit-cpu-parts-definition-
function.html
[Accessed 29 03 2020].