Are generative design tools creative? A characterisation of tools throughout the design process

Abstract

Generative design is now a core capability of today's CAD systems enabling broader exploration of the design space. This paper presents a desktop review of 15 tools with the aim of characterising where the tools are used in the design process and to what extent they can be considered creative.The results showed that, despite their increasing capabilities, GD tools are unable to exhibit all phases of creativity and cannot be applied in all stages of the design process. This highlights the criticality of the human in design providing opportunities for GD to augment designer creativity.

Full text

DESIGN METHODS AND TOOLS 683
INTERNATIONAL DESIGN CONFERENCE – DESIGN 2024
https://doi.org/10.1017/pds.2024.71
Are generative design tools creative? A characterisation of tools
throughout the design process
Owen Rahmat Peckham , Ben Hicks and Mark Goudswaard
University of Bristol, United Kingdom
owen.peckham.2019@bristol.ac.uk
Abstract
Generative design is now a core capability of today’s CAD systems enabling broader exploration of the design
space. This paper presents a desktop review of 15 tools with the aim of characterising where the tools are used
in the design process and to what extent they can be considered creative.The results showed that, despite their
increasing capabilities, GD tools are unable to exhibit all phases of creativity and cannot be applied in all
stages of the design process. This highlights the criticality of the human in design providing opportunities for
GD to augment designer creativity.
Keywords: creative process, creativity, design process, generative AI, generative design
1. Introduction
Generative design (GD) systems were first conceptualised by Mitchell (1975) for architectural design.
In the following two decades very few publications gave the topic any further attention (Caetano,
Santos and Leitão, 2020). In recent years, development has returned to the field with an increasing
publication rate - 288 published with the phrase ‘generative design’ in 2022, up from 80 in 2017
(Scopus, 2023) - and the release of several tools for both academia and industry such as Fusion 360
GD suite (AutoDesk, 2023) and CATIA GD Engineering (Caetano, Santos and Leitão, 2020; Dassault
Systemes, 2020).
With increased computing capability, in particular the development of cloud computing services, and
the maturation of Additive Manufacturing (AM) capable of constructing complex geometries, GD tools
are becoming increasingly feasible for practical use in engineering. Fusion 360’s GD tool, which can
perform either structural or fluid path studies and is aimed specifically at mechanical engineering
applications, is a good example of this. An example of the tool’s use for generation of a mobile phone
stand structure can be seen below in Figure 1a and Figure 1b.
Released in 2023 OpenAI’s DALL-E 3 and similar generative AI tools also demonstrate capability to
generate designs (Bing, 2023). DALL-E 3 can generate images directly from text prompts, an example
of which can be seen in a design generated with the prompt ‘mobile phone stand’, shown in Figure 1c
1
.
Engineering GD tools are principally utilised for their ability to widen the search into different solution
types. This ‘design space exploration’ increases the ability to present new and high-performing solutions
to a design problem. GD tools facilitate this design exploration in several ways. Many of the specialist
engineering GD tools are built upon an optimisation framework that enable them to explore the design
space as they search for the ‘best performing’ solution.
1
As an aside, note how the interpretation of ‘mobile’ has been applied to the ‘stand’ as well as the ‘phone’.
https://doi.org/10.1017/pds.2024.71 Published online by Cambridge University Press

684 DESIGN METHODS AND TOOLS
Figure 1. Examples showing usage of generative design tools
GD tools also facilitate exploration through their lack of bias. When humans tackle design problems
they necessarily begin with preconceived notions of what the solution may look like due to their
upbringing and cultural drives (Sawyer, 2006). Specialist engineering GD tools do not have these
constraints
2
and instead are constrained by the characteristics of the output of their algorithm.
Finally, the ability of GD tools to speed-up design iterations can also act to increase the design space
search as it enables designers to consider a greater quantity of options. Generative design tools can
create several highly developed solutions within hours or minutes. This capability allows designers to
compress the iteration period and focus time elsewhere.
As the use of GD tools becomes widespread, several aspects of how they can contribute to a
designer’s workflow remains unclear. First, where can these tools be utilised within the design process?
Experienced designers develop procedures and habits as they tackle problems. Therefore, if they are to
adopt GD tools as part of their workflow, they require the knowledge of exactly how and where the
tools can add value. This characterisation forms part of this paper’s contribution.
Second, can GD tools be considered creative? During progression through the design process there are
multiple instances where creative input is required to ensure a workable solution (Cross, 1982; Howard,
Culley and Dekoninck, 2007). As such, it is necessary to understand what aspects (if any) of creativity
these tools display so that designers can decide how to utilise them most effectively. The authors can
find no other work that focus upon these areas at the time of writing. This paper continues by examining
the related work on GD tools, the design process and creativity (Section 2). This is followed by an
explanation of the desktop review method used to evaluate the tools (Section 3). Next, a mapping of the
tools with respect to the design process and the creativity aspects that they display is presented (Section
4). The implications of the mappings are discussed (Section 5). The paper then concludes with the main
findings from the investigation (Section 6).
2. Related work
This section provides the chosen definition of GD and justification for the use of design and creativity
process models. Process models are considered from the literature and the most appropriate is
selected.
2
This is unlikely to be true for tools that have been trained such as DALL-E 3 as they are likely to have absorbed
human bias during their ‘learning’ process.
(a) Mobile phone stand generative
design study geometry setup with
Fusion 360
(b) Result of Fusion 360
generative design study
(c) A design created using
DALL-E 3 with the prompt
‘mobile phone stand’
https://doi.org/10.1017/pds.2024.71 Published online by Cambridge University Press

DESIGN METHODS AND TOOLS 685
2.1. Generative design tools
There is not an agreed-upon definition of GD. For the purposes of this investigation, the definition
presented by Caetano et al. (2020) will be used: “GD is a [computational] design approach that uses
algorithms to generate designs”, where generative is the “capacity to produce or create something”.
Other definitions such as those given by AutoDesk (2023), Bernal et al. (2015) and Oxman (2008),
make further distinctions between algorithms based upon the required input and number of generated
outputs. This study, however, will not consider these distinctions as they narrow the scope of the
investigation by limiting the number of eligible tools. Functionally, the implication is that for the
purposes of this investigation, no categorical distinction is made between topology optimisation and
GD.
In engineering a ‘design’ is a blueprint that defines a machine system through some means. The stage
of the design process likely dictates the form this design takes, with earlier stages relying on abstract
natural language and later stages focussed upon less ambiguous dimensioned drawings.
Several methods are used to achieve GD capability. Traditionally, these have mainly been
L-systems, Cellular Automata, Genetic Algorithms, Shape Grammars and Evolutionary Design
(Caldas, 2008; Singh and Gu, 2012; Abdelmohsen, 2013; Caetano, Santos and Leitão, 2020;
Starodubcev et al., 2023). Fundamentally, with regards to application to structural generation, these
methods primarily operate by adding and removing cells of a meshed domain in accordance to defined
stress criteria. The consideration that an AI tool such as DALL-E 3 can operate as generative design
systems adds additional methods to this list. Generative AI tools function by using various types of
algorithmic Artificial Neural Networks, maintaining consistency with the chosen GD definition
(Ramesh et al., 2022).
2.2. The design process
At present, GD is most often discussed in the context of structural optimisation (e.g. Díaz et al., 2021;
Wang, Zhang and Bernard, 2021; Barda et al., 2023). This necessitates that the discussion focuses upon
the finalisation of the design process. It is generally agreed, however, that creativity is most obviously
displayed earlier in the design process (Snider, 2014; Wodehouse and Casakin, 2022). This study
therefore endeavoured to include and characterise tools throughout the design process to create a
comprehensive view of the level of creativity of all GD tools.
To situate the GD tools within the engineering design process, requirements were placed upon the model
used in the study. The model needed to reflect development from idea to artefact, be applicable to the
design of machine systems, and balance generalisation with fidelity within the design stages. The Design
Council (2006) double-diamond framework, Gero and Kannengiesser (2004) FBS framework and Pahl
and Beitz (2007) were compared to determine which was most appropriate given these requirements.
The Pahl and Beitz (2007) model was found to be most applicable. It centres around the four stages of
Task Clarification, Concept, Embodiment and Detail which broadly explain how designers can progress
from their initial ideas through to a complete solution. During Task Clarification the designer defines
the problem by considering objectives, constraints, stakeholder requirements and feasibility. In the
Concept stage designers search for solution principles that can tackle the problem. In the Embodiment
stage this solution principle is refined into an architecture. Last, in the Detail stage the architecture is
finalised into a specific structure that performs suitably. This model was considered appropriate due to
its 4-stage fidelity within the problem-solving process and common usage describing technical and
machine system design.
In addition to adopting a process model, it is necessary to define the interpretation of some terms for the
purpose of the study. Aeroplane landing gear will be used as an example case. The ‘problem definition’
considers the need for a cushioning system that dampens landing sufficiently to prevent damage to the
aeroplane and its occupants whilst allowing a low friction interface with the ground for quick take-offs.
A ‘solution principle’ is the type of mechanism settled upon following the concept stage; this might be
wheels as opposed to landing skids. The ‘architecture’ is the configuration achieved; the wheels are
arranged into three sets, one under the nose and one under each wing. The ‘final structure’ is the
complete design; each set will contain two wheels of 0.5m diameter. In the Concept through to Detail
stages the designs can be represented in a variety of ways - from natural language to dimensioned
https://doi.org/10.1017/pds.2024.71 Published online by Cambridge University Press

686 DESIGN METHODS AND TOOLS
drawings - with the abstraction decreasing and precision increasing as the process advances as discussed
above in Section 2.1.
2.3. Creativity
Creativity is regarded as an essential attribute for a successful designer, allowing them to tackle
problems in a variety of new, interesting, and applicable ways (Askland, Ostwald and Williams, 2010;
Wodehouse and Casakin, 2022). As such, if GD tools are to increasingly complement human designers
in their work, a gauge of their creative ability is necessary. Sarkar and Chakrabarti (2007) and Sawyer
(2006) explain that the two necessary conditions for creativity are novelty and appropriateness.
However, satisfying these two conditions is not straightforward. It requires both real world evaluation
and either re-purposing of existing technology or invention of new technology (Sawyer, 2006).
The subjective nature of creativity makes direct and quantitative measurement impossible when
considering a range of problems due to its cultural and experiential influences (Amabile, 1996; Sawyer,
2006). An analogue for creativity measurement was therefore employed. To do this, a gauge of creativity
was taken by mapping the tools’ processes to a model of the creative process. This allowed insight into
the level of creativity of the tools whilst also serving as a guide for the abilities that the tools would need
to gain to be considered creative.
For their study into the intersection between the design and creative processes, Howard et al. (2007)
compare nineteen creative process models and suggest common phases that most of the models include.
These phases were chosen as an appropriate creative process model for use in this paper due to
the rigorousness of Howard’s study and its wide acceptance: it has been cited 90 times for use in other
studies. Howard finds that the phases of the creative process are:
• Analysis - Problem interpretation
• Generation - Incubation and synthesis of ideas (preferably many)
• Evaluation - Selection of the most appropriate solution from all the available options
• Implementation - The process of maturing the solution to use/ application via commit,
manufacture or deployment
Howard’s inclusion of Analysis and Evaluation within the model serve to ensure that only intentional
design serves as creativity. Without recognition of these phases creativity can be identified in random
physical processes: is a glacier creative in its carving of a valley?
3. Methodology
The Methodology section begins by explaining the desktop review method used to characterise the
design tools. The placement of tools within the design and creative processes is then demonstrated with
examples.
3.1. Desktop review
To map the processes exhibited by the tools against the chosen design process and creative process
models, a desktop review was conducted of each of the tools. For this, an initial list of 28 tools was
found from an internet search of design tools beginning with the terms: ‘generative design tools’ and
‘engineering generative design tools’. These 28 tools form a non-exhaustive list. Further investigation
narrowed the list to 15 tools that comprised sufficient related information for the authors to make a full
characterisation as outlined in Section 3.2 and 3.3. The desktop review was then conducted for these 15
tools to assess their inputs, operation, and outputs. To do this, academic literature, promotional material,
and case studies were considered amongst other sources. This process is illustrated in Figure 2.
The 15 tools selected for review were: TRIZ (Gadd, 2011; Petrov, 2023), Functional Analysis (Little,
Wood and McAdams, 1997), Brainstorming (Putman and Paulus, 2009), ChatGPT (Filippi, 2023),
Morphological Analysis (Arciszewski, 2018), DALL-E (Bing, 2023; Brisco, Hay and Dhami, 2023),
CATIA GDE (Dassault Systemes, 2020), Inspire (Lim and Wong, 2018), Ansys (Ansys, 2021), Abaqus
(Johnsen, 2013), Sulis (Altair, 2023; Das et al., 2023), Siemens NX (Menninger, Berroth and Jacobs,
2022; Siemens, 2023a), AutoDesk Fusion 360 (AutoDesk, 2023), Simulink (MathWorks, 2023) and
Amesim (Siemens, 2023b). The four italicised early-stage design tools in this list are not computational
https://doi.org/10.1017/pds.2024.71 Published online by Cambridge University Press

DESIGN METHODS AND TOOLS 687
methods. Whilst computational tools exist to aid in the application of these tools, they are fundamentally
applied by human designers. As such these tools do not fit satisfy our definition for GD. It was decided
that these tools should be included in the review as they provide a comparison against the generative AI
tools and aid in understanding of the methodology.
Figure 2. The desktop review process
3.2. Design process characterisation
The input and output of each tool was used to characterise the design stage that the tools operate within.
Table 1 demonstrates the inputs and outputs used to classify each tool as operating in a specific design
stage. Simcenter's Amesim is a good example of how these inputs and outputs were used to characterise
and map the tool against the design process (Siemens, 2023b). Amesim is a design tool that aims to aid
the design of multi-physics mechatronic systems. Designers interface with Amesim's tools and solvers
using block diagrams where components are selected from a library. Multi-physics systems can be built
by assembling these components. Here the input of the generative tool is the architecture established by
component selection and connection by the human designer. The generative algorithm aims to search
and optimise component parameters. The output is a specific parameter set that gives the design a
finalised structure. The generative capability of Amesim therefore sits within the Detail design stage as
it takes an architecture as input and outputs a finalised structure.
Table 1. Attributes used to characterise the tools within the design process
Input
Output
Task Clarification
Mess or Obstacle
Defined problem
Concept
Problem statement
Solution principle
Embodiment
Solution principle
Architecture
Detail
Architecture
Finalised structure
3.3. Creative process characterisation
The tools' inputs, algorithms and outputs were also used to characterise their capability with respect to
Howard's phases of creativity. The inputs and algorithms required by the Analysis phase are unclear;
this will be discussed further in Section 5. The output of the Analysis phase is a defined problem.
The Generation phase of creativity can occur at all stages of the design process (Howard, Culley and
Dekoninck, 2007). The input to the Generation phase can therefore be in the form of an obstacle,
problem statement, solution principle or architecture, as shown in Table 1. A tool that supports
Generation should create and synthesise designs, conforming to the earlier definition of GD. The tool
should then output a set of solution principles, architectures, or finalised structures.
https://doi.org/10.1017/pds.2024.71 Published online by Cambridge University Press

688 DESIGN METHODS AND TOOLS
The Evaluation phase reviews the design following Generation. The input to this phase could therefore
be a solution principle, architecture, or finalised structure. The design could be scrutinised with physics-
based methods such as Finite Element Analysis (FEA) to assess its feasibility. A tool that provides
Evaluation should output a measure of the viability of the chosen solution.
To support Implementation a tool needs to progress the finalised design to a prototype or product. To
do this fully in an engineering machine system context would require the integration of a process such
as AM into the toolchain. Partial exhibition of Implementation could be achieved with integration of a
method of Computer Aided Manufacture (CAM) such as slicing for AM.
4. Tool mapping
The rows of the flowchart in Figure 3 show the mapping of the 15 tools against the position they hold
within the design process. The GD tools can be split into two categories, the early-stage general purpose
tools and the late-stage specialist tools. The placement of the four traditional and non-generative tools
provides context to the framing of the generative Concept and Embodiment stage tools.
The columns of the flowchart in Figure 3 demonstrate the phases of Howard's creativity process that
each of the design tools cover. In this figure the generative capability of the tools is represented by the
boxes containing the tool's name in bold, the normal font names represent capability that is feasible
within the wider software package if augmented by a human designer, and the non-generative tools are
distinguished with italicised names.
ChatGPT and DALL-E are considered to perform some of the steps of Task Clarification as part of their
generative capability. Both are capable of Generation of a problem definition however require a prompt
including an obstacle and an objective. For example, ChatGPT presents solutions that necessitate it
having generated problem definitions. When prompted ‘There is a closed door in front of me. I need
something the other side of the door.’, it answers with a list of potential solutions including ‘Use a key’
and ‘Check if it's unlocked’. This shows the tool is capable of Generation in the Task Clarification stage.
Meanwhile, the requirement for a clearly stated obstacle demonstrates that ChatGPT is incapable of
Analysis in the Task Clarification stage. Similarly, neither tool performs any form of Evaluation upon
the derived problem definitions before they are Implemented.
The Embodiment capability of Amesim and Simulink is entirely human designer driven. It requires a
human designer to select, organise and connect the components to create a system. These tools' GD
capability sits within the Detail design stage as explained in Section 3.2. Amesim and Simulink are
primarily multi-physics simulation tools. As such, they have capacity for Evaluation of the designs they
produce in multiple contexts given the input of a human designer. Ansys, Inspire, Abaqus and Sulis are
all capable of Evaluation for the same reason.
CATIA, Fusion 360 and Siemens NX are all CAD/ CAM tools. This extended capability, along with
their simulation packages, enables these tools to demonstrate aspects of Implementation and Evaluation
when operated by a human designer. The CAM capability does not provide full Implementation but is
the first step toward a physical product as discussed in Section 3.3.
5. Discussion
The Analysis phase of the Task Clarification stage of Pahl and Beitz (2007) engineering design process
model stands notably empty in Figure 3. In this stage designers consider that there is an obstacle to an
objective. Howard et al. (2007) that the creative process is present within each stage of the design
process, acting as the mechanism by which the input of the stage becomes the output: The Analysis
phase of Task Clarification encompasses the identification of the obstacle. The Generation phase
synthesises problem definitions. In Evaluation the definitions are assessed for appropriateness.
Finally, in the Implementation stage the problem is defined in a way that allows a Concept design
stage tool to interpret it. These phases of Task Clarification stand separately from the Analysis phase
of the later design stages where defined problems are interpreted. This study argues that no design
tools exist for performing Analysis or Evaluation at the Task Clarification stage. This is discussed
further below.
https://doi.org/10.1017/pds.2024.71 Published online by Cambridge University Press

DESIGN METHODS AND TOOLS 689
Figure 3. Illustration of the phases of Howard's creative process that the tools exhibit within
each stage of the design process. Location of bold tool names indicate position of the tools core
generative capability, normal names show tool functionality that is feasible in conjunction with
a human designer and italicised names represent the non-generative tools
The generative AI design tools capable of operating in the early Concept and Embodiment design stages
are general purpose tools built upon natural language processing. The flexibility afforded by the text
representation of designs allows the tools to operate across broad design principles with high abstraction
and ambiguity (Filippi, 2023). This generality and therefore lack of specialised physics models prevents
these tools from operating in the Detail design stage, however.
The Detail design stage contains specialist engineering tools that cannot support earlier stages of the
design process. The specificity at this late stage requires the accuracy of high-fidelity physics models.
Like general purpose tools, the specialised nature of engineering tools prevents them from providing the
flexibility required by earlier design stages due to their application of an existing rigid knowledge base.
These extremes of general and specialised tools highlight a gap unfilled by GD tools and represents a
trade-off between the depth of insights a tool can provide and the breadth of its applications.
Ordinarily industry focuses upon development of high readiness level tools whilst academia works upon
more novel approaches to problems. Over the past couple of years, however, the development of
generative tools has been driven principally by large technology companies, with OpenAI a standout
example. In contrast, academic progress has been focused particularly upon specialist engineering
problems (Mountstephens and Teo, 2020; Acosta-Zazueta, Alcaide-Marzal and Diego-Más, 2021). This
https://doi.org/10.1017/pds.2024.71 Published online by Cambridge University Press

690 DESIGN METHODS AND TOOLS
approach has meant that methods developed by academia contribute mainly to development of Detail
stage generation but leaves the earlier design stages untouched.
Given the current capability of design tools, the human designer remains with two important roles to
play in creative engineering design. First, none of the tools can operate within the Analysis or Evaluation
phases of the Task Clarification design stage. Pfeifer and Bongard (2006) suggest that the abilities
needed to exhibit Analysis in Task Clarification requires immersion within an unconstrained
environment such as the physical world and a pressure to mature. Current GD tools are heavily reliant
upon constraints to guide their synthesis. The specialist engineering tools are unable to exhibit Task
Clarification because of their inability to generalise and approach the problem from multiple directions.
The Evaluation phase of the Task Clarification and Concept stages are similarly unpopulated because
such tools would require a limitless yet precise knowledge base to give accurate predictions of feasibility
given only an abstract input. These weaknesses demonstrate the need for a human designer to identify
a problem, research it, then define its scope, constraining the problem for the tool.
Figure 3 demonstrates that engineering tool software packages are capable of more aspects of the
creative process than Generation. These capabilities are currently disparate, however. This opens the
second avenue of contribution from the human designer. The wider Fusion 360 software package is
capable of GD, FEA and CAM but needs a designer to link the three appropriately. Increased integration
of tool capabilities through linked cloud-based software packages facilitates rapid and easy use of the
various functionalities, enabling increased design creativity. Model Based Systems Engineering is an
example of linking these capabilities but is mainly human actuated (Madni and Sievers, 2018).
The emergence of generative AI tools and their use for engineering design suggests future development
opportunities for systems less reliant on human guidance. Specifically, the ability to train generative AI
on large knowledge bases could in future allow gaps to be bridged from early to late-stage design.
Using the phases of the creative process that the tools' exhibit allows a gauge to be taken of tool
creativity. Importantly, none of the tools are capable of functioning generatively in all the four phases
of the creative process and each rely on human designers in some areas. This need for human designer
guidance implies that it is more appropriate to think of GD tools as aiding the creative process and
enhancing the designer's creativity, rather than as creative themselves. For a tool to demonstrate
creativity it would need to be capable of autonomously progressing through each of the phases of the
creative process within one or more design stages.
6. Conclusion
This study aimed to establish the scope current generative design tools across the design process and
how creative the design tools are. A desktop review was performed of 15 design tools, including 4 non-
generative tools for comparison. The review considered the tools' inputs, the nature of their algorithm,
and their output. Using these characteristics, the capability of each tool was mapped to the Pahl and
Beitz (2007) engineering design process. It was found that the generative capability of all the
engineering specific tools sat within the Detail design stage and only generative AI tools could operate
in the earlier concept and embodiment design stages in a generative capacity. This highlights the gap
between the use of flexible general-purpose tools which can define problems and specialist engineering
tools better suited to the rigid application of knowledge bases for late in the design process.
To explore the creativity of the tools they were mapped also against the four phases of Howard et al.
(2007)'s creativity process. Here it was found that the specialist engineering tools exhibited just the
Generation phase autonomously with only the natural language based generative AI tools supporting
other phases of the creative process. The implication is that generative design tools are only capable of
assisting the creativity of the designer as they do not exhibit all phases of the creative process. The
emergence of generative AI tools suggests that the aspects of creativity the tools display will likely grow
however the intervention and guidance of humans with remain necessary for creative design for the
foreseeable future.
Acknowledgements
EPSRC DTP 2020-2021 University of Bristol Grant Ref EP/T517872/1
https://doi.org/10.1017/pds.2024.71 Published online by Cambridge University Press

DESIGN METHODS AND TOOLS 691
References
Abdelmohsen, S.M. (2013) ‘Reconfiguring Architectural Space Using Generative Design and Digital Fabrication:
A Project Based Course’, in Proceedings of the XVII Conference of the Iberoamerican Society of Digital
Graphics - SIGraDi: Knowledge-based Design. Valparaiso, Chile: Editora Edgard Blücher, pp. 391–395.
Available at: https://doi.org/10.5151/despro-sigradi2013-0074.
Acosta-Zazueta, G., Alcaide-Marzal, J. and Diego-Más, J.A. (2021) ‘GENERATIVE DESIGN SOFTWARE
AND DIFFERENT APPROACHES’.
Altair (2023) 3D Printing Generative Design Software | Altair Sulis, Default. Available at: https://altair.com/sulis
(Accessed: 13 October 2023).
Amabile, T.M. (1996) Creativity in context: Update to ‘The Social Psychology of Creativity.’ Boulder, CO, US:
Westview Press (Creativity in context: Update to ‘The Social Psychology of Creativity.’), pp. xviii, 317.
Ansys (2021) Ansys Topology Optimization | Lightweighting & Shape Optimization. Available at:
https://www.ansys.com/applications/topology-optimization (Accessed: 1 November 2023).
Arciszewski, T. (2018) ‘Morphological Analysis in Inventive Engineering’, Technological Forecasting and Social
Change, 126, pp. 92–101. Available at: https://doi.org/10.1016/j.techfore.2017.10.013.
Askland, H.H., Ostwald, M. and Williams, A. (2010) ‘Changing Conceptualisations of Creativity in Design’.
AutoDesk (2023) Fusion 360 Generative Design Extension. Available at: https://www.autodesk.co.uk/products/
fusion-360/generative-design-extension (Accessed: 2 November 2023).
Barda, A. et al. (2023) ‘Generative Design of Sheet Metal Structures’, ACM Transactions on Graphics, 42(4), p.
116:1-116:13. Available at: https://doi.org/10.1145/3592444.
Bernal, M., Haymaker, J.R. and Eastman, C. (2015) ‘On the role of computational support for designers in action’,
Design Studies, 41, pp. 163–182. Available at: https://doi.org/10.1016/j.destud.2015.08.001.
Bing (2023) Bing Image Creater - powered by Dall-E3, Bing. Available at: https://www.bing.com/images/
create/image-creater/646a1181ae664496a8f83e5bb8607d35 (Accessed: 6 November 2023).
Brisco, R., Hay, L. and Dhami, S. (2023) ‘EXPLORING THE ROLE OF TEXT-TO-IMAGE AI IN CONCEPT
GENERATION’, Proceedings of the Design Society, 3, pp. 1835–1844. Available at:
https://doi.org/10.1017/pds.2023.184.
Caetano, I., Santos, L. and Leitão, A. (2020) ‘Computational design in architecture: Defining parametric,
generative, and algorithmic design’, Frontiers of Architectural Research, 9(2), pp. 287–300. Available at:
https://doi.org/10.1016/j.foar.2019.12.008.
Caldas, L. (2008) ‘Generation of energy-efficient architecture solutions applying GENE_ARCH: An evolution-
based generative design system’, Advanced Engineering Informatics, 22(1), pp. 59–70. Available at:
https://doi.org/10.1016/j.aei.2007.08.012.
Cross, N. (1982) ‘Designerly ways of knowing’, DESIGN STUDIES, 3(4).
Das, S. et al. (2023) ‘A case study: Analyzing the structure and optimizing the design of the aero handlebar
assembly for the bicycle used in Japan’s Olympic games’, Engineering Reports, p. e12783. Available at:
https://doi.org/10.1002/eng2.12783.
Dassault Systemes (2020) Generative Design Engineering. Available at: https://events.3ds.com/generative-design-
engineering (Accessed: 1 November 2023).
Design Council (2006) Framework for Innovation - Design Council. Available at:
https://www.designcouncil.org.uk/our-resources/framework-for-innovation/ (Accessed: 7 November 2023).
Díaz, G. et al. (2021) ‘Applications of generative design in structural engineering Aplicaciones del diseño
generativo en la ingeniería estructural’, 36.
Filippi, S. (2023) ‘Measuring the Impact of ChatGPT on Fostering Concept Generation in Innovative Product
Design’, Electronics, 12(16), p. 3535. Available at: https://doi.org/10.3390/electronics12163535.
Gadd, K. (2011) TRIZ for engineers: enabling inventive problem solving. Chichester: Wiley.
Gero, J.S. and Kannengiesser, U. (2004) ‘The situated function–behaviour–structure framework’, Design Studies,
25(4), pp. 373–391. Available at: https://doi.org/10.1016/j.destud.2003.10.010.
Howard, T., Culley, S. and Dekoninck, E. (2007) ‘CREATIVITY IN THE ENGINEERING DESIGN PROCESS’,
in. ICED07.
Johnsen, S. (2013) Structural Topology Optimization. NTNU Norway. Available at: https://ntnuopen.ntnu.no/
ntnu-xmlui/handle/11250/241794.
Lim, S.T. and Wong, T.T. (2018) ‘Unleash the potential of additive manufacturing with topology optimization’,
in. 6TH INTERNATIONAL CONFERENCE ON PRODUCTION, ENERGY AND RELIABILITY 2018:
World Engineering Science & Technology Congress (ESTCON), Kuala Lumpur, Malaysia, p. 040007.
Available at: https://doi.org/10.1063/1.5075571.
https://doi.org/10.1017/pds.2024.71 Published online by Cambridge University Press

692 DESIGN METHODS AND TOOLS
Little, A.D., Wood, K.L. and McAdams, D.A. (1997) ‘Functional Analysis: A Fundamental Empirical Study for
Reverse Engineering, Benchmarking and Redesign’, in Volume 3: 9th International Design Theory and
Methodology Conference. ASME 1997 Design Engineering Technical Conferences, Sacramento, California,
USA: American Society of Mechanical Engineers, p. V003T30A015. Available at:
https://doi.org/10.1115/DETC97/DTM-3879.
Madni, A.M. and Sievers, M. (2018) ‘Model-based systems engineering: Motivation, current status, and research
opportunities’. Available at: https://doi.org/10.1002/sys.21438.
MathWorks (2023) Simulink - Simulation and Model-Based Design. Available at: https://uk.mathworks.com/
products/simulink.html (Accessed: 10 November 2023).
Menninger, B., Berroth, J. and Jacobs, G. (2022) ‘Development of System Alternatives using Generative
Engineering’, in 2022 IEEE International Symposium on Systems Engineering (ISSE), Vienna, Austria: IEEE,
pp. 1–6. Available at: https://doi.org/10.1109/ISSE54508.2022.10005535.
Mitchell, W.J. (1975) ‘The Theoretical Foundation of Computer-Aided Architectural Design’.
Mountstephens, J. and Teo, J. (2020) ‘Progress and Challenges in Generative Product Design: A Review of
Systems’, Computers, 9(4), p. 80. Available at: https://doi.org/10.3390/computers9040080.
Oxman, R. (2008) ‘Digital architecture as a challenge for design pedagogy: theory, knowledge, models and
medium’, Design Studies, 29(2), pp. 99–120. Available at: https://doi.org/10.1016/j.destud.2007.12.003.
Pahl, G. and Beitz, W. (2007) Engineering design: a systematic approach. 3rd ed. Edited by J. Feldhusen and K.H.
Grote. London: Springer.
Petrov, V. (2023) Talented Thinking: TRIZ. Cham: Springer International Publishing. Available at:
https://doi.org/10.1007/978-3-031-15505-5.
Pfeifer, R. and Bongard, J. (2006) How the body shapes the way we think. The MIT Press.
Putman, V.L. and Paulus, P.B. (2009) ‘Brainstorming, Brainstorming Rules and Decision Making’, The Journal
of Creative Behavior, 43(1), pp. 29–40. Available at: https://doi.org/10.1002/j.2162-6057.2009.tb01304.x.
Ramesh, A. et al. (2022) ‘Hierarchical Text-Conditional Image Generation with CLIP Latents’. arXiv. Available
at: http://arxiv.org/abs/2204.06125 (Accessed: 7 November 2023).
Sarkar, P. and Chakrabarti, A. (2007) ‘DEVELOPMENT OF A METHOD FOR ASSESSING DESIGN
CREATIVITY’.
Sawyer, R.K. (2006) Explaining creativity: the science of human innovation. Oxford ; New York: Oxford
University Press.
Scopus (2023) Scopus - Document search. Available at: https://www.scopus.com/search/form.uri?
display=basic#basic (Accessed: 24 November 2023).
Siemens (2023a) Siemens NX Generative Design Software, Siemens Expert Partner. Available at:
https://www.ddsplm.com/products/nx-cad-generative-design/ (Accessed: 1 November 2023).
Siemens (2023b) Simcenter Amesim, Siemens Digital Industries Software. Available at:
https://plm.sw.siemens.com/en-US/simcenter/systems-simulation/amesim/ (Accessed: 30 October 2023).
Singh, V. and Gu, N. (2012) ‘Towards an integrated generative design framework’, Design Studies, 33(2), pp.
185–207. Available at: https://doi.org/10.1016/j.destud.2011.06.001.
Snider, C.M. (2014) Characterising the Creative Behaviour of Designers within the Late-Stage Engineering Design
Process.
Starodubcev, N.O. et al. (2023) ‘Generative Design of Physical Objects using Modular Framework’, Engineering
Applications of Artificial Intelligence, 119, p. 105715. Available at: https://doi.org/10.1016/
j.engappai.2022.105715.
Wang, Z., Zhang, Y. and Bernard, A. (2021) ‘A constructive solid geometry-based generative design method for
additive manufacturing’, Additive Manufacturing, 41, p. 101952. Available at: https://doi.org/10.1016/
j.addma.2021.101952.
Wodehouse, A. and Casakin, H. (2022) ‘Design Creativity in Architecture and Engineering’, Buildings, 12(10), p.
1552. Available at: https://doi.org/10.3390/buildings12101552.
https://doi.org/10.1017/pds.2024.71 Published online by Cambridge University Press