Decolonial AI: Art Science as a Regenerative Tool

Sophia Davies, Laura Ghebali Boukhris

1. Introduction

The role of design within technology innovation is becoming more widely understood, with

human-centered methodologies and tools increasingly applied within the product

development lifecycle of organizations; however, a framework for applying broader artistic

practices to technology innovation remains largely undeveloped.

Our research focuses on one specific innovation, that of artificial intelligence (AI). Artificial

comes from the Latin artificialis meaning “of or belonging to art”, and so the place of art in AI has always beckoned. Indeed, this innovation is becoming more and more present in the

products used in our everyday life, whether the mobile interfaces with which we carry along

with us or the machines we use to increase productivity or efficiency at work. Indeed, AI is

integrated from the very early creation stage up to the diffusion and deployment of it. The

rapid rate by which development is accelerating requires that we question ourselves on the

nature of AI as it occupies an increasingly larger portion of our everyday live. What and

whom is behind AI’s development? How has it been developed and according to which

view? Why would we accept to outsource our intelligence to a machine and for what

purpose? These are all questions that arise from the basic fact that AI holds human

authorship, a design process that assumes an identity, a logic, and one that should surely

reflect a minimum of common thoughts from the individuals it impacts, not just those of its

author group.

As AI continues to intercept and assume ever increasing space in our everyday lives, we

believe there is a significant urgency to develop a systematic approach to decolonization –

and one that depends critically on the value of lateral thinking born of the artistic practice to

mediate a necessary shift from human-in-the-loop to society-in-the-loop AI development.

Whilst there exists research that points to the integration of creativity to concrete

organizational product development, research exploring the direct relationship between art

and technology innovation remains limited. Furthermore, the subject of decolonial AI is still

very much an emerging field. Our research intends to address the gap between artistic

critical practice and technology innovation, whilst at the same time building on nascent

thinking in the field of decolonial AI to identify art’s place in more sustainable technology

innovation.

(2. Full literature review and 3. methodology included in the full article - to be requested directly to the author) 2. Literature review

4 Art Science for Decolonial AI

Through the synthesis of the research results, we seek to codify the findings into the

requirements of a new design ecology for AI development, one that can be leveraged as a

tool for regenerative innovation in decolonisation contexts. We speculate on a critical framework for Art Science to operationalise and scale the decolonization of AI within industry, outlining cognitive, strategic, and organizational constructs. The Art Science framework for Decolonial AI is made up of: (i) a new mental model; (ii) new design systems (or “systems of decolonisation”), and; (iii) an ecosystem model for operationalisation within the org (see Appendix 1).

4.1 Towards a Critical Practice of Art Science

To apply Art Science as a critical practice in the work of decolonial AI, we need a new mental

model with which to see the world. We have inspired our synthesis of ideas and themes from

our methodological field, which is summarized through the below word cloud. Consequently,

we concluded that the research identified three roots of the critical practice of Art Science,

namely interaction, reasoning and knowledge, and elevated cognition, for which each has a

set of branches by which to navigate the world.

Table 5 : Word cloud as a synthesis to the methodological field

Interaction

The research identified interaction as a key characteristic of the critical practice of Art

Science, in particular the modes of perception and treatment of artefact in respect of

material origin.

Perception

As a first step and to set the stage of the research, one needs to reflect on what the use of AI

implies in terms of relationships between the human body and technology itself. The

philosophy of interaction between technology and human is underlined by the fact that the

human body is a living organism, implying evolution as well as sensitivity to context.

Differently to a machine or any technology, the human body reacts to the surrounding

environment, and does so with a sensitive fragility. We are not “matter” nor a “fitted” elementthat remains unchanged like writing set in stone, contrary to a machine. But, and perhaps paradoxically, we are developing technologies that evolve; we are creating machine learning algorithms in aspiration of the self-organising complexity attributed to the evolution of human

beings – whether doing so consciously or unconsciously.

More specifically, we often look at technology as a way to displace humans so that

automation and learning will “give back” time to individuals. But this conception implies

considering us two playing on the same scene, but this is not the case at all, we are not in

any manner competing with this technology. We built technology and the current

environment is a unique opportunity for humans to continue to master the machine as

humans always comes before technology. And even if we were to consider ourselves on the

same level, for example by considering us as a simple data set, we would need to define us

as a data set that is “in movement” because that is the essence of human being - to

continuously evolve. Yet, the fact that we are trying to somehow replicate the human through

machine while we are developing technology and AI causes one particular issue that caught

our attention: what type of individuals are we now trying to replicate? Which vision is

currently entitled into the innovation of AI and how inclusive it is? These questions clearly

recentralize the stakes of integrating artists into innovation development to ensure human -

in a broader sense - are well represented and inclusive.

A topic of reflection was how our perception of the objective reality – that which ‘just is’,

without ambiguity – depends on our frame of reference, and informs our lived experience.

Through this, artists are reconceiving design as an ontological act that we might term ‘design

being’; an embodied design practice wherein there exists a dialogue between form and

matter, between creator and environment. The progression from ‘thinking’ to ‘being’ is

meditated through experiencing structural and temporal relationships, inhabiting the

objective reality and subjective experience of the present lived-through moment within and

beyond the constraints of technology.

This tendency for art to exhibit new states of being – perhaps ‘ecologies of being’ – has

important implications for the enmeshed reality of physical and digital matter i.e., data. It

presents us with an experiential frame of reference for how AI, though absent of the state or

quality of being, intercepts the ecologies we inhabit and generates stimuli that go on to

become part of our lived experience. This is reinforced in the literature of Oxman (2016),

where she poses the question; “when practicing art, is perhaps what truly counts less the art

form and

more one’s (way of) being?” and has been much the focus of artist and researcher Sougwen

Chung (2021) through explorations on “ecologies of becoming with”.

Artifact & Origin

But to consider this clear differentiation between the natural essence of human versus

technology, it is important to reflect on the core elements of a technology. As mentioned by

some of our interviewees, the technology itself is a material artefact made up by humans.

Technology innovations are the products of art and human is always the chief of art - no

matter the timing, no matter the evolution of the subject matter. Having said that, we can

clearly understand the conception of technological matter as a prolongation of human body,

highly mentioned in the literature as an extension of human capabilities (Brey 2000), and

what artists Sougwen Chung (2020) terms a form of “planetary sensing”. Other researchers

specify this idea by explaining that these artefacts amplify human capacities - either that of

the physical body, or our mental devices ; (Hutchins 1999 Clark 2004) and it contributes to

the identification of the peculiar position that technology occupies today facing human with

the rapid development of AI.

On a more granular conception, most artist practicians share the priority of respecting the

matter as such. Specifically, if there is no questioning around the important recentralization

of technology around human, this is not to forget the added value brought by working with

the matter itself. This latter has an identity to be respected, and the idea of the “matter being

stubborn” came back in a few discussions. A bit like if we were to expect a rock to be flexible

or a leaf to be hard, each technology as a matter linked to physical components that

represent constraints for the human to work with. Indeed, this constraint often forces

individuals to overcome the barriers and have their initial ideas of art creation evolve while

considering the matter we are manipulating. And this is a beauty of it and a skill uphold by

artists: working around constraints and finding ways to overcome the unexpected. This is

almost a philosophy around the respect of the matter that is offered to us on the planet, and

this notion is important as well to consider when participating to the development of AI: how

do we respect the matter of AI, which is to say the datasets? How do you use it in a way that

doesn’t take off their initial meaning through the process of de-contextualization that is often

required during the structuration of it?

This problematic relates to the design of AI and the functioning that emerges from its

conception. Either being natural language processing, machine learning, voice or face

recognition or any other category of AI, the link between raw materials (data set) and the

utility of the final product is directly made by the way it is designed. But as explained by one

AI expert interviewer, the role of a designers is intentional, while the art isn’t at all. The focus

of Art is to give a particular interest to process while the role of design is to ensure a specific

utility of the product itself. We therefore understand through these talks one contribution of

artistic practices to the development of AI, which is to ensure the full respect of the dataset

as subject matter and how it is encoded/structured to respect its origin.

Reasoning & Knowledge

The combined epistemic practice of art, philosophy and science was central to the value

proposition of Art Science to decolonial AI, the research exploring key themes across

reasoning and knowledge.

Reasoning

The research synthesis covered multiple modes of ‘thinking’ – from creative and critical

thinking, through to design and art thinking – and, whilst each was attributed its specialist

distinctions, a common emphasis was placed on the verb ‘thinking’ indicating the importance

of reasoning in all derivatives of artistic practice.

It was also a point of reflection how the fundamentals of thinking are being deeply changed

through technology, so far as encouraging a radical rethinking of thinking itself (L.

Brabandere, 2022). Emphasis was placed on the need for new mental models to

accommodate advancements in technology, particularly in evolving our definition of

anomalies – which in the case of AI takes on new urgency. The relationship between

thinking and decision-making was explored, the bounded rationality that characterises

behaviour-based decision-making and how it manifests through how we design technologies

such as AI, not just how we use them. This suggests that the economic considerations

attributed largely to the fields of engineering and design in the literature (Oxman, 2016), do

have their place in Art Science and extends notions of utility and intentional change –

specifically, behavioural – to its work.

The role of Art Science emerges around creating a culture of inquiry and “a climate of doubt”

(Iny & Brabandere, 2013), one of critical questioning aimed at probing the dimensions of

ambiguity, was one of importance. When liberated from preconceived assumptions and

hypotheses, we can consider alternative hypotheses through inductive thinking and go

beyond the deductive thinking of existing theories. We might say inductive thinking is

“out-of-the-box” thinking, but it is in fact the act of thinking in entirely new ones (Brabandere

& Iny, 2013). The stimulus for inductive thinking resides in how we go about questioning the

world around us, the way we channel our curiosity with the intentionality to expand our

imaginative capacity (Fuller & Reeves, 2021). Art Science can help generate new, inductive

pathways through which to explore decolonisation – perhaps a decolonial realisation of

Buckminster Fuller’s “synergetics”, the study of spatial complexity to encourage new kinds of

lateral thinking in solving problems (A. Edmonson, 1987).

Iny & Brabandere (2013) describe the foundations of predictive thinking as residing majorly

in deduction, in contrast with prospective thinking which rests more heavily on induction.

Since a large portion of AI is focused on the development of predictive models, it is

interesting to consider the elevated importance of prospective reasoning – perhaps in

analysing multiple futures – and reinforces the role of Art Science as being a regenerative

tool. Through this, we see that Art Science finds synergy with Speculative Design – and

Critical Design more generally – by virtue of its prospective nature. Reflecting on the notion

of designed realities, it is interesting to think about how the taxonomy of futures could be

mediated by Art Science, in particular pathways to preferable (e.g., decolonised AI) futures.

In exploring the motives that drive the work of art and science, reference was made to how

both similarly strive to change our perception of the world. To call upon the work of Art

Science as reorganised perception pushes art beyond its individual work in expressing

subjective experience. It might suggest Art Science to be a practice that strives to create a

continuum of perception, superposing all contexts in a sort of objective whole. This plurality

– over universality – holds important implications for its use in decolonial approaches to AI.

Further, the research reinforced the delineation of Art Science thinking from Design thinking,

and contrasted their distinct focus areas – for example: process versus product;

project-creator fit versus product-market fit; expression versus action; perception versus

production, and; vision and philosophy versus products and services.

An interesting contrast was made between the work of art versus innovation, specifically how

the latter affects a change in objective reality, not just our (subjective) perception of it.

Therein emerges an important distinction of Art Science from Art, perhaps by virtue of its

origins in critical practice; Art Science seeks to change a reality – a colonised one – and in

his way ebbs and flows with innovation.

In relation to this is the act of bisociation (Koestler, 1964) – the linking of two associative

contexts to disclose hidden axioms and ‘latent logic’ – and its role in poesis, particularly in

the creation of new bodies of knowledge. This appears to be the philosophical isomer of

Kauffman’s (2004) adjacent possible, and through this lens we appreciate the pervasiveness

of human thought in innovation and the elevated role of Art Science as a tool through which

to “reinforce epistemic practice" (Milan and Van der Velden, 2016). This way of thinking,

maximising the utility of all existing bodies of knowledge, holds significance for decolonial AI. By examining combinations of cross-domain knowledge, Art Science creates space for the

pluralism within which real-world AI exists and finds potential application in cross-context

data discovery and analysis.

Knowledge

In creating meaning through Art, we consider a semantic stance on the references from

which knowledge stems.

It is through the aggregation of data we often form new bodies of knowledge, but it was

emphasised that too often we lose the individual contexts from which these bodies originate

and afford little attention to the systemic morphology through which ‘artificial knowing’

emanates. Art Science, perhaps through its roots in the design science of Buckminster Fuller

(1975) in the literature, was identified as a practice that could help preserve the

context-profile of the data, and how it changes over time by virtue of aggregation

transformation and so on and so forth. Reference was made to thick data, and its role in the

big data lifecycle – data reflecting for example social or personal contexts and attributes –

and elevated the importance of new, context-preserving approaches to data method

associated with AI.

The role of ethical and epistemic values in the formulation of new knowledge was explored,

within which Art Science could provide a more diversified and contextualised approach to

data and AI, scaling novel human-in-the-loop approaches – such as ‘data alchemy’, for

example (Candelon et al., 2021) – more towards our aspirational society-in-the-loop future.

The research was suggestive that through the Art Science practice, we can evolve more

systematic approaches to overcoming the data colonialism inherent of the accelerating data

economy – that concerned with the intentional exploitation of data for economic growth

(Mohamed et al., 2020) – by better tracing the colonial attributes of data aggregated and

transformed across systems within society. This could bring about the ‘new data

epistemologies’ referenced by Milan and Van der Velden (2016), those that equally privilege

humans and nature. We use the term data ecology to denote one such epistemology that

could be mediated by Art Science, one focused on a critical infrastructure that advocates for

ecological traceability and transparency of data, and respects data as a material resource

that should not be exploited for economic purposes alone. This has important considerations

for how we rethink approaches to such principles as data lineage and digital provenance,

and Art Science, through its epistemic devices, can help encode more ethical semantics for

how we govern and manage data.

This places Art Science central in moving development processes away from extractive and

exploitative practices – such as data mining – and towards regenerative practices. This also

encourages us to challenge our pre-conceptions of economies of scale and speculate on

what ecologies of scale would imply.

The research illuminated how our understanding of the interrelationship between knowledge

building and lived experience is being challenged through the disruption of technology on our perceived field of experience (Brabendere, 2022), namely through increasingly complex

interdependence with data. From the way data is aggregated and analysed, through to how it

is actuated and interfaced with through digital products and services, data shapes our lived

experience and henceforth how we go about generating new knowledge. Building on this, a further consideration was made on how power asymmetries are positively reinforced through uneven degrees of design agency in data and AI products, meaning we see the metropoles referenced by Mohamed et al. (2020) dominate the knowledge networks associated with AI research and development. The synthesis indicated that Art Science could help establish modes through which to decentralise AI knowledge, particularly by increasing

the breadth and diversity of those versed and engaged in the process.

Explored was also the ability of Art Science to increase the effectiveness of tacit knowledge

transfer – that is, knowledge gained from experience – through new dissemination

techniques, namely interaction and participatory performance, was identified as being

potentially instrumental to the field of decolonial AI. It encourages the demo and

simulation-based development practices of AI to be challenged and for new techniques of

community engagement and real-world impact analysis to better overcome algorithmic

exploitation – for example, in the beta-testing environment as indicated by Mohamed et al.

(2020). Since tacit knowledge is rooted in lived experience, context, and value systems, the

ability to collect and transfer this information at scale, whilst at the same time preserving its

structure, is key to decolonial approaches.

Art Science was a field thought to be associated with manifold models of knowledge, thinking

through the space of all possible representations of a subject or topic – based on existing

bodies of knowledge – to generate new, plausible encounters (Reeves & Fuller, 2021). We

might think of this conceptually as a ‘latent space’ of knowledge – owing to the term used for

neural networks – one that acts like an inductive field for the aforementioned bissociative

thinking and catalyses effective recombination of cross-domain knowledge.

Colwyn (2019) references a similar concept of the “liminal realm” as being the space

between bodies of knowledge where the boundary of one “system of knowing” ends and

another begins. That in occupying the liminal space – inhabiting this sense of between-ness

– of subjective experience and the objective reality, art manifests itself (J. Colwyn, 2019).

The idea of an objective knowing (J.Colwyn, 2019) and ‘systems of knowing’, and where

“other mediums of expression thrive” in the ‘voids between structures of established thinking

where truly compelling questions emerge’. Catalytic combination of two forms – Art and

Science – come together to evoke new forms of knowledge.

In this way, Art Science might lend itself – to quote the words of artist Refik Anadol (2020) –

as a “prediction engine for humanity” through which to “remember the future”, thereby

reinforcing itself as a speculative tool in exploring the manifold of all possible, probable, and

plausible knowledge.

Elevating Cognition

The research signified the role of Art Science in elevating the cognitive capacity of individuals

and teams with the field of AI, identifying illumination and paradoxical thinking as two modes through which it creates these cognitive shifts.

Illumination

But why would Artists hold such a role ? This questioning brings us back to the literature

linking art and innovation. When asks as such in the semi guided interview, our corpus

reveals the idea that art is an added value because it doesn’t fit in any classical schemas.

Indeed, art doesn’t look for definition, neither classification – it is unbounded. The role of

artist is to create troubles, to initiate thought that comes from different frameworks of

meaning and that would not have come up if people were to stay in their usual environment.

Art pushed people out of their comfort zone to constantly challenge what is considered as

“crystal clear” or “confirmed”. As such, It acts as an illuminator that depicts ‘alternativism’

and plural futures where people don’t see any or more than one and this discovery of new

routes often reshape expectation. Thus, artists open the mind and avoid any kind of mind

closeness. They open possibilities to make sure all angles are approached, from all

disciplines, all characters, because they do not belong to any. They are considered as

free-mind and this essence provides freedom that, in a world of technology, ensures ethics

and inclusivity while acting as a guide to develop new ways of working.

Paradox

In a broader scope, another specificity of the artistic approach is the creation of awareness

around a certain number of paradoxes that are inevitably important when we talk about

technology innovation, and specifically AI. The research indicated an important critical tension bought about by confronting such paradoxes and putting them to use to spur on new

thinking.

The first one to think about is related to the fact that Art puts his interest into the process of

creation rather the final product itself and its usage. Which turns down the notion of

economics and performance to a more granular thinking on how we create innovation.

People could argue that no matter the process, the most important aspect to look at is to

make sure the final product matches the expected product that creator had in mind. In the

case of AI specifically, we understand that the way dataset and algorithm are constructed

have a direct impact on the way AI will shape the everyday life of numerous amounts of

people. When data scientists are left alone in this creation process, there is a danger and

fear to only focus on the useability of a final product and to forget that the use is indeed,

directly impacted by the way we constructed it because we are talking about technology, and

not human. This differentiation can partly be explained by the fact that expression prevails action within artistic approaches. Expression of feelings, sensitivity, or expression of what a specific process involved in terms of behaviour, as opposed to action that concentrate eyes only into the movement itself, at one point in time, which is often link to a specific result. The

two-paradox explained above rekindle the essence of the difference between a subject and

an object matter and can be linked as well to the other related paradox of the notion of

enhancement versus optimization. Indeed, the AI goal is to optimize in a continuous way the

level of specificity that it can answer. Differently, human value if to constantly try to enhance

their ongoing actions and results: the way of thinking and reasoning if highly different, which

is also another argument making the displacement of human by technology erroneous.

The other paradox brought by artists is the notion of evolution in comparison to the

fossilization of thought form. In a word and in addition to the explanation of the evolutionary

aspect of humans, one challenge in the AI creation and development if related to the fact

that the design impacts the way we format users, meaning other people brain. Through the

high development of use of an AI, we format the brain and fossilize it by clustering the

imagination and the interaction that we can have with the machine.

Besides, part of the intent behind the creation of an algorithm is to create pattern of answers,

which means in some way a simplification of the reality to enable automated answer. This

notion is indeed opposite to the notion of complexity that more specifically artistic

approaches bring. The goal is to understand the reality and its different layers on a the most

granular possible way in order to avoid misinterpretation or ellipsis. Complexity versus

simplification represents another paradox that opposes human and innovation AI.

The position of artists to exact the “double vision” spoken of in the research of Mohamed et

al. (2020) was a key point of discussion, with potential to confront power asymmetries

through explorative techniques leveraging paradoxical mental models. For example, the

lens of “metropoles vs periphery” could be leveraged by Art Science to codify new modes of

recentralisation.

Collective Engagement

By bringing this whole range of paradox into the discussion, Art and artistic approaches bring

a collective consciousness to the stakeholders involved in AI creation, that is meant to

ensure a sustainable development for our coming years and coming generation. It helps the

society as a whole and on a more granular dimension, different group of people to re-create

the consciousness of the right and sane relationship that human should maintain while

interacting with technology and creating new ones that impact millions of people’s routine

life. It develops social imagination that prevents dangers to happen while seeing the world

with open eyes without being refrained from any mental thinking or company’s code or

culture. Opening the barriers help group to imagine alternative scenarios – worst or better –

that have the benefits to elevate people vision.

4.2 Systems of Decolonisation

To give shape to the Art Science field and understand what it means to apply such a field in

the context of decolonial AI, we begin to speculate on a series of interventions through which

decolonisation can be enacted across the AI lifecycle and forward our conceptions of

“society-in-the-loop”. Notably, the application of Art Science interventions to decolonial AI

builds on the work of Mohamed et al. (2020), referenced in the literature review.

Based on our series of interviews and our workshop, we have identified data curation, artistic

dissemination, bisociative data discovery, regenerative adversity, participatory research, and

safeguarded augmentation as potential mechanisms of decolonial intervention, identifying

the most relevant intervention points for their application across the AI lifecycle.

These interventions are intended to indicate potential testbeds for developing a set of

methods for the Art Science practice within organisations and serve as areas for further

debate and research.

Let us deep dive on each intervention mechanism of Art Science to explore how they can

facilitate new pathways towards decolonial AI.

Data Curation

Intervention points: Business Framing; Data Collection; Model Validation; Deployment

The first mechanism that we have identified through our research as a relevant mechanism

for decolonial AI is the artistic curatorial practice, and how it can reimagine technology-led

data curation practices in development teams today.

The term curation has evolved over time, but Larousse (2014) defines the practice as

specifying that it aims to select and share the most relevant content of the web for a specific

purpose. In the academic literature, the Centre for Digital Curation in England describes this

practice as "an activity of managing and promoting the use of data from its creation, to

ensure its adequacy with contemporary objectives and its availability for discovery and

reuse" (Westbrooks, 2008; Lord & Macdonald, 2003). It can be held here that curation is

presented as a practice carried out with the aim of achieving a specific objective. What is

more, this mission must make it possible to reuse the information. Finally, curation is also

understood as an added value: "A process of maintaining value addition to a reliable body of

information, for purposes of current and future use" (Greenberg, White, Carrier, & Scherle,

2009). This is a similar aspect to the field of art; the curator accompanies the works of added

value to propel them. It does not offer a creative activity in its own right.

In our context, we understand the importance that data curation can have to answer our

problematics. Specifically, curators could help “integrate more contextual awareness of

human conceptions of fairness” and safety through a data-to-value alignment and the

realignment of ‘implicit value-systems’ to those associated with ethics (Mohamed et al.,

2020). It helps people enlarge their view and horizon on the topic and it could help

overcome what has been referred to as a specification problem of AI (Mohamed et al.,

2020), – that is to say the contrast between the ideal specification and that revealed when

deployed in the real-world.

Having said that and to be more concrete, curatorial practices could be considered as a

solution to our problematics while being embedded at the feature extraction stage of model

development, working closely with data scientists to examine e.g., confounding variables. It

is also a way to make sure the design of AI is more inclusive, while integrating a broader

and more diverse datasets. In this context, selecting the data will be a more continuous

process

and thus, less categorical in nature that will bring more diversity. Curatorial practices would

objectively confront homogeneity and power asymmetries to power up a process of AI

creation that has explicit value. As mentioned by Mohamed et al. (2020), there is a need to

recognise implicit value systems and it seems that data curation could participate to that

goal. At the end, the data curation could be considered as a curatorial technique that

participate to help

structure protective mechanisms against exploitative or extractive data practices that also

value the contextualization of them.

Looking at the AI lifecycle in a more exhaustive way, from data collection to deployment and

going through model validation, integrating the technique of curation has additional benefits;

it could detail the origin of the data through data lineage, but also the way it is used and

structured through the work of profiling. The selection and categorization aspects that imply

the curational technique is useful at every stage of the AI creation considering that the raw

material is about data. The more you curate, the more you select, the more you source and

the less value you lose from the data, because you know how to “treat” data and how to

collect it.

However, the integration of curation into AI creation is not a random practice to learn on the

go. Validating this assumption of the need of curator to be integrated into the design of AI

would require curating training for specific profiles, where the curators would test and

validate datasets thanks to their curatorial practice.

All in all, the application of Art Science here could encourage the heterogenous engineering

practice proposed by Mohamed et al. (2020) and more context-aware technical development

processes. It could also encourage a wider size of collaboration in data curation, and

reinforce

Artistic Dissemination

Intervention points: Data Collection; Model development; Deployment; Use

The role of dissemination within the Art Science practice was identified a potential mode to

encourage decolonial techniques in the AI development and deployment scenarios.

At the data collection stage, for example, Art Science could be a device through which to

make explicit the representation of knowledge assumed within a dataset and communicate

effectively what is and is not included. Within model development, use cases for Art Science

were identified in the UX design of development interfaces, specifically the interaction design

to notify developers in real-time of potential input or parameterisation biases. It was

emphasised that to design such features effectively, productively intercepting and modulating

algorithmic development processes, would require the skills of communication, mediation,

and behaviour science promising of Art Science.

Further, going beyond the internal development teams of organisations and into real

deployment scenarios, Art Science was considered a means by which to create more

effective model schemas to improve the interpretability – e.g., model accuracy, applicable

contexts, data origins – and make more explicit their limitations in informing more ethical

use safeguarding by the right traceability mechanisms within real-world applications. For

example, Google developed the ‘Model Cards Framework’ to better communicate the

accuracy of

existing models and algorithms to end users (Mohamed et al., 2020), and Art Science could

help build on such frameworks and ensure their scalability.

Bisociative Data Discovery

Intervention points: Business Framing; Data Collection; Model Validation & Testing

The other mechanism we would like to highlight as a potential mechanism for decolonial AI is

what we call bisociative data discovery.

As mentioned by Ahmed and Fuge (2018), “Bisociative knowledge discovery” is an approach

that combines knowledge elements from two or more incompatible domains with the goal to

generate creative solutions and insight. This notion is inspired by Koestler (1964) who

designed himself a model of creativity under the notion of bisociation. Ahmed and Fuge

landed the idea of bisociation to what they call a bisociative networks, a knowledge graph

that can “capture conceptual similarities between ideas and helps designers find creative

links within that network”.

In the context of decolonial AI, bisociative knowledge discovery could be extended to a

bisociative data discovery process, combining datasets from two or more ‘incompatible’

contexts to generate new aggregations that could disclose ‘hidden realities’ and provide

adjacent decolonial pathways. Further, the practice of exploratory data analysis (EDA) could

be broadened via Art Science to include such bisociative analysis techniques and better

discover patterns and anomalies of significance that exhibit colonial characteristics.

Draw on the intersection of Art Science in Speculative Design: leveraging ‘Signal scanning’

in the scenario space of the speculative design practice, with the artist as a “critical antenna”

in having orthogonal fields and viewpoints.

(Re)Generative Adversity

Intervention points: Business Framing; Model Evaluation

Another mechanism revealed by Art Science that could be beneficial to reach a

decolonization of AI is what we termed “regenerative adversity”.

It finds itself a metaphor through the neural nets of Generate Adversarial Networks (GANs),

which can be thought of as representing a minimal society of two – what B. Arcas (2019)

refers to as a sort of “social imagination”. The generator and discriminator could be likened

to the role of creator and critic – “is it competition, is it cooperation, is there a difference

between the two" (Arcas, 2019) – and through Art Science we see “the reflexive work of

criticism” noted by Mohamed et al. (2020) a key enabler in creating the sort of decolonial

learning loops within organisations that catalyse more regenerative AI design and

development practices, e.g., more continuous model evaluation processes.

Our research illustrates the value proposition of artists for creating this sort of ‘critical

tension’; one that maximises the innovation capacity of collective groups of people –

mirroring “productive adversity” B. Arcas (2019) refers to of GANs – and evolves them along

new decolonial pathways towards society-in-loop outcomes.

Participatory Research

Intervention points: All

Furthermore, the practice of participatory design embodied by Art Science swerves as a

decolonial intervention to encourage more “society-in-the-loop” development practices.

More specifically, we believe turning the design of AI into a participatory action would help

creating a community that makes the AI more authentic and inclusive. To build on this idea,

Mohamed et al. (2020) explains on his words how such co-development practices can instil

‘algorithmic accountability’ through participatory action research, with design and

development

‘driven by the agency, self-confidence and self-ownership of the communities they work for’

This idea introduces the paradox between self-expression and collective affiliation and

pushes towards a “society-in-the-loop” ambition for AI development. Artists are well

positioned to encourage such co-design for algorithmic interventions alongside the

communities they are deployed in.

In addition, considering participatory design pushes us to consider its impacts in terms of

helping to encode different modes for decentralisation, redistributing power away from the

‘colonial monopoles’ and into the ‘peripheries’. Namely, there is an undoubted need to

include more of the society in the design of AI which brings us to think that the Art Science

mechanism help, by engaging communities and incentivising equal participation across

societies, in the recentralisation of knowledge.

Coming back to the colonial monopoles and the peripheries and to deep dive this metaphor,

one could argue that given the fact that metropoles often took learning from periphery, the

participatory design involved by the integration of Artist in AI decolonization would enable the establishment of a reserve tutelage between centre and periphery that contribute even more to AI design that better reflects the different faces of the society.

Safeguarded Augmentation

Intervention points: Business Problem Framing

A final mechanism identified for the application of Art Science as a decolonial intervention,

was that relating to measured and ethical judgement on the right augmentation model of

humans via AI. Specifically, informing decisions on the degree by which AI either replicates

or augments human capabilities, and ensuring these decisions are safeguarded by the right

guiding principles.

The research indicated that companies must understand the different combinations of human and AI to tackle business problems and choose the right combination model for a given context (Candelon et al., 2022). For this to happen, sufficient space needs to be given for divergence of thinking before jumping to solutions. Through Art Science as an ethical

calibration tool, we might safeguard business framing decisions around both the ‘use

case–AI fit’ and the ‘degree of AI augmentation’ of human processes to better ensure

decolonial practices in the early phases of the AI lifecycle.

Thus, Art Science comes into the place of helping to identify where on the human-AI

intelligence spectrum a specific use case resides and developing critical frameworks to guide

the design decisions around its application to the business context. This has important

implications on ways of working within the organization and may serve to better elevate

employees and encourage reinforcing infrastructure such which can act to encourage the

behavioural changes needed in the organization to effectively enable Art Science.

4.3 The Art Science Ecosystem

This section looks at entity and infrastructure of an extended Art Science ecosystem.

Our focus here is on organisational constructs and required integration structures.

This section looks at how we can operationalize the mechanism of Art Science and apply it

within organization to make sure artistic approaches are embedded to the design and

development of AI.

More specifically, we have identified a spectrum of possible archetypes for embedding Art

Science into the organisation, within and beyond AI/data science units.

(A) Experimental

construct via an ArtSci Lab

(B) Systematic construct via existing Innovation structures

(C) Decentralised function via embedded teams

Let’s deep dive on those different archetypes to understand the operationalization of the

intervention mechanisms outlined previously.

Organizational Archetypes

***

(A) Experimental construct via ‘ArtSci Lab’

In this archetype, Art Science is positioned as a dedicated function e.g., ArtSci Lab or Hub, a

specialised capability built within a specific team independent of functional units similar to

product, engineering or marketing. Such dedicated functions have had notable success in

the inception of Innovation and Design entities in largely immature or traditional

organisational settings.

The focus of the Art Science Lab is to explore new and different applications – often

experimental and/or innovative – and often focuses on self-contained projects. It is catalytic

and divergent in nature, where the artist encourages the organisation to adopt increasingly

experimental processes ‘from which they will extract know-how by inference and

abstraction’.

This archetype strongly advocates for reciprocal approaches, where functional experts like

data scientists or Machine Learning engineers contribute their expertise to Art Science-led

initiatives.

It also depends on the availability of resources, such as open data, tooling, and interaction

based setups incl. physical installation spaces. It was suggested that it would be a natural

extension of the computational ‘sandbox’ environments, into the physical which could be site

specific. This could be accelerated by more hybrid approaches leveraging Web 3.0 and

Metaverse. Further, the rotation of artists, working on portfolio of pilot use cases over a

period of one to two years. We need to mobilize a set of things (object, space, sign) for us to have an interaction with the reality. We ensure people plug their thoughts into reality, but we don't reach organizational level.

Whereby the artist’s potential to serve as a critical antenna for society and culture is

amplified, able to explore what might matter or generate value and use future literacy to

speculate and inspire on anticipatory constructs that seek to strengthen the practice of

increasingly leveraged disciplines such as Design Futures and Speculative Design. In this

vein, the Art Science Lab construct can serve to augment strategic foresight and adaptive

capabilities. Extending typical hackathon initiatives, or the concept of growth hacking, to

something more aligned to a type of “ecological hacking”.

This type of organisational setup would build upon the Artist in Residence approach from

leading organisations e.g., S+T+ARTs, and similarly the fellowship programs led by research

and development labs e.g., MIT Media Labs.

(B) Systematic construct via Innovation

Differently, in this archetype, Art Science is positioned as an embedded practice within

broader innovation functions and/or capabilities (whether distributed or decentralised), with

associated methods and principles forming innovation pathways.

There exist two modes within which Art Science could add value; either in product innovation contexts, whereby Art Science methods are embedded in design, development, and deployment teams, or in process innovation contexts, where Art Science informs more of a strategic influence on product and growth agendas and actions itself through continuous

improvement initiatives. Artists act as principal decisions makers for AI products and Process champions for a more continuous AI development process

Whether product or growth hacking, Art Science can be embedded in existing ways of

working and delivery mechanisms. The challenge would be conflation with the practice of

design more broadly – despite certain points of commonality, the distinctions remain critical

and should be considered.

(C) Decentralized functional Art Science

In this archetype, Art Science is decentralised and distinct by each functional unit, the

objective being for each function to develop its own Art Science capability. Delivery is

managed at the level of each functional unit and their existing processes. Here, artistic

approaches of the Art Science will be more specialised to each function’s unique needs and

aspirations, and able to adopt different maturity levels in accordance with this. The Art

residency is a good example of this operationalization, but the risk of this approach is that it

perpetuates existing silos.

In the medium or long term, there would eventually be a need to extend beyond established

structures e.g., residencies and consider how to industrialize and scale them across

industries and sectors, without losing the continuity of cognitive, strategic, and

organizational constructs.

Challenges

Having said that, no matter the chosen archetypes, and according to our research, we have

identified that these operational modes need to tackle at least three major challenges,

namely: aligning purpose and common goals; accommodating different processes and ways

of working, and; quantifying value and impact. Let’s deep dive on each of these challenges.

When talking about purpose and common goals, we relate to the tension that arises from

simultaneously trying to fulfill individual goals – artists alongside roles within the functions,

e.g. product, engineering, marketing – and a collective objective. In the case of Art Science,

individual purpose is majorly centred on exploring the unknown and disclosing ‘hidden

realities’, with value generation an indirect consequence of the pursuit of meaning. This is in

contrast to strategic and portfolio-based objectives of initiatives and projects driven from the

business that centre around what is quantifiable – often related to resilience, efficiency, and

effectiveness.

The second challenge is related to the processes and ways of working. Indeed, the artistic

rhythm exists in stark contrast to the Agile cadence and “demo” culture intrinsic to product

and software development, and artists cannot be expected to produce in lockstep on such

timelines. There is without any doubt a lack of a common language and distinct ways of

working that prevents artistic approaches to be integrated into organization. Finding points of convergence with the incremental workflow will be critical however, to modulate the design, development and deployment processes, and require that we challenge concepts around ‘minimum viable’ and ‘fail-fast’ paradigms. The required distinct processes require ‘handling discoveries and the unexpected from different points of view’ (Henchoz et al., 2019).

Lastly, the expectations on quantifying the value and impact of Art Science to the

organisation need to be clearly managed as it is not as directly measurable as other such

innovation practices as growth or design, for example. Equally, Art Science will be

challenged to find more tangible means by which to assert its value in strategic contexts,

and its success will depend on equal compromise with the organisation to find the right

aperture through which to measure the quality and ‘regenerative impact’ over time.

Solutions to Challenges

To overcome these challenges, only the awareness of them is essential but still not enough.

Our research methodology blending interviews and workshops have enabled us to point out

critical success factors for effective embedment, and for cultivating a culture for co-creation

between artists and development teams.

Purpose

The first element would be to Identify the typology of problems to be solved for, and within

this exercise, the initial framing and inception was identified as being fundamental to

success. Broader and more amorphous problem spaces being favoured over more narrow

framed/vertical ones, business problems that already have ‘shape’ and pre-conceived

hypotheses limit the danger of missing the right purpose through the high involvement into

ambitious projects that have been identified as fertile ground for such initiatives.

Setting clear expectations of roles and outlining the collective objective at kick-off is a must

need in these contexts, meaning that the goals of art and engineering co-creation must be

clearly specified. Potential for disruption, social impact, economic impact are known to be

strong common drivers and great examples of potential common goals to create interaction

between such a two diverse teams. Yet, preserving the integrity of each discipline is still

challenge. There is a definite need to navigate the boundaries of ‘cultural domains and

identities’, forwarding individual purpose within each contributor, whilst simultaneously

advancing the collective vision.

In addition, and still in this perspective of preserving integrity, the setting of a shared

vocabulary helps artists to grasp the technology, preserve their own creativity and share their

visions. Having a facilitator to ‘transduce’ the languages within the room and convert these

into shared conceptual understanding can be a great first step. The role of artist itself was

said to embody this facilitatory role – able to create a ‘meta-language’ in the support of

common goals and break down the barriers across disciplines. However, in the context of

combined output, it was suggested this needs to be neutral person. In the case of the

S+T+ARTS residency program (Henchoz et al., 2019), success was largely attributed to the

role of mediator and their instrumental role in key decision points of the process –

“especially to support the initial definition of the technology challenge and to run the

inception meeting”.

Process

In parallel, solving the process challenges is directly linked to the importance of the

availability of the right resources, in the form of expertise, infrastructure such as tools and

environment(s), and artefacts. As an illustration, to provide an environment primed for co

creation, S+T+ARTS recognised the merit of creating an ‘in-between space’ inclusive of

heterogenous identities and equal access to shared artefacts (Henchoz et al., 2019). Having

a loosely integrated methodology to guide the interaction and relationships between artist

and functional experts is also a good partial solution as it induces a dynamic flow throughout

the co-creative process, maintaining the creativity conductivity through which Art Science

thrives.

The process is still modulated to some degree by an iterative workflow to help better predict

needs and showcase outputs on a periodic cadence.

Yet, the role of the artist within the wider team needs to be well defined with clearly set

expectations, adopting a similarly egalitarian approach as any other functional role. Artist in

Residences are considered through the lens of mutual development, whereby artists

enhance their strategic fungibility and functional experts expand their critical thinking and

creativity quotient. The exchange must be from both perspectives, questioning not only what

the artist contributes to functional domains, but equally what engineers and product designers contribute to the artist. All in all, having a clear set of expectations the integration of artists within the organization would help overcome the internal tension between creation and critique embodied in individual architypes and encourage a more ‘intrapreneurial’ inception within organisations.

Finally, establishing the right incentives model for participation in Art Science practices was

identified as being fundamental to mobilising resources within the organisation. This requires

new incentives models to garner the contributions of functional experts across the

organisation to volunteering their practice to Art Science collaborations.

Product

And to finish with, a few solutions scenarios have been identified through our research

methodology to tackle the product challenge defined earlier. Among a few, the most

important ones to highlight is about the importance of implementing lighthouse projects early on as mechanisms for building trust and appreciation and recognition of the impact imparted from Art Science collaborations.

Additionally, there is a need for mediating the transition from proof-of-value and towards

‘proof of-meaning’ by observing contextual significance and an ongoing evolution of the

project beyond its original inception and first material results. On a more general perspective, t was noted that valuing the process of creation, rather than the product alone,

can be inspirational for all stakeholders, inviting diverse groups to co-create, share cultural

reflections and experience, building new collective dialogues and foundations of trust. This

develops and embeds a deep respect and conceptual understanding for the artistic practice,

in particular as a valid practice in the production of knowledge.

Finally, one mechanism through which the output space can be expanded, is by the

establishment of new learning architectures in the organisation through Art Science. The role

of artist and related bodies of work as a connective medium across all levels of the

organisation can help establish learning loops in a sort of ‘create and learn’ approach. There

should be a focus on designing critical feedback loops to enable this sort of continuous

learning and its self-reinforcement. By creating productive adversity, the mediation of

individual to organisational knowledge occurs in a self-organised way.

All in all, the three dimensions of purpose, process and products are inherent to any type of

organizational architecture and serve as a reusable framework for any such structure of a

new field of expertise within an existing one, such as, in our context, Art Science within AI

development.

5 Conclusion

The objective of our research was to explore the requirements of a critical framework for the

application of Art Science as a regenerative tool in the field of AI – regenerative by way of

facilitating decolonial interventions within the design, development, and deployment of AI.

Our research speculated on a critical framework comprising cognitive, strategic, and

organizational structures required to operationalise and scale the application of Art Science

to decolonial AI, and explored new mental models, design systems and organizational

structures.

Three key elements were proposed for the mental models associated with the practice of Art

Science: interaction, knowledge and reasoning, and elevated cognition, synthesizing the

branches of cognition within each as indicated by our findings.

We identified a series of potential interventions through which new decolonial pathways can

be created via Art Science, namely the mechanisms of data curation, artistic dissemination,

bisociative data discovery, regenerative adversity, participatory research, and safeguarded

augmentation, each acting to forward our conceptions of society-in-the-loop development

practices. A critical next step in the research is to prototype these interventions as they apply

to the AI lifecycle, and methodise the mechanisms explored.

Finally, we analyzed the key challenges to be overcome in embedding Art Science into

organizations, proposing solutions around purpose, process, and product, and exploring

three archetypes for operationalising the practice across and within functional domains of

existing organisations.

Conclusively through our research, we see the elevated role of Art Science as a mutually

reinforcing symbiosis of art and data science, encouraging new data epistemologies in AI

whilst simultaneously creating new decolonial techno-cultural grammars in society through

art. In this way, Art Science has the capacity to advocate a shift away from the extractive and

exploitative processes of the data economy and towards the regenerative processes of what

we term a new ‘data ecology’. Equally, through the creation of new works of art, Art Science

can democratise the narrative of AI and reinforce its role as a generative tool in society.

Acknowledgements

We would like to thank all persons who have accepted to give their time to be interviewed

and who shared their expertise with us.

We thank ArtTech for their collective collaboration. We also extend thanks to Fanny Potier for

her facilitation of select interviews during the research, and for providing her feedback on the

workshop structure.

We are grateful to artists Joanne Bloch, Refik Anadol and Sougwen Chung for their

collaborations in recent years.

Disclaimer: Any opinions in this article represent the personal views of the authors and do

not necessarily reflect those of their respective organisations.

APPENDIX

References

Agre, P. E. (1997). Lessons learned in trying to reform AI. Social science, technical

systems, and cooperative work: Beyond the Great Divide, 131.

Ahmed, F., & Fuge, M. (2018). Creative exploration using topic-based bisociative

networks. Design Science, 4.

B. Arcas, (2019) Social Intelligence, NeurIPS, 2019

https://www.youtube.com/watch?v=uyUbGatPKpI

Bhatti, Y. A. (2012). What is frugal, what is innovation? Towards a theory of frugal innovation.

Towards a Theory of Frugal Innovation (February 1, 2012).

Björgvinsson, E., Ehn, P., & Hillgren, P. A. (2010, November). Participatory design

and" democratizing innovation". In Proceedings of the 11th Biennial participatory

design conference (pp. 41-50).

Bloch, J., & Verchère, C. (2019). How art places climate change at the heart of

technological innovation.

Brown, T., & Katz, B. (2011). Change by design. Journal of product innovation

management, 28(3), 381-383.)

Carlsson, B. and Stankiewicz, R. (1995): “On the nature, function and composition of

technological systems”.

Carlsson, B. (ed.): Technological Systems and Economic Performance: The Case of Factory

Automation. Kluwer Academic Publishers. Dordrecht, pp. 21-56.

Chung, S. (2002). Building a national innovation system through regional innovation

systems. Technovation, 22(8), 485-491.

Clark, K. B. (1985). The interaction of design hierarchies and market concepts

in technological evolution. Research policy, 14(5), 235-251.

Coeckelbergh, M. (2018). The art, poetics, and grammar of technological innovation

as practice, process, and performance. AI & SOCIETY, 33(4), 501-510.

Gano, G. (2015). Starting with Universe: Buckminster Fuller's Design Science Now. Futures,

70, 56-64.

Edquist, C. and Johnson, B. (1997): “Institutions and Organizations in Systems of

Innovation”. In: Edquist, C. (ed): Systems of Innovation: Technologies, Institutions and

Organizations. Pinter Publishers. London, pp. 41-63.

E. Peters, AIxDesign (2019) “Imagining narratives for preferable AI futures”

https://medium.com/aixdesign/imagining-narratives-for-preferable-ai-futures-185a0a99b2e

9

F. Candelon, T. Reichert, S. Duranton, and R.C di Carlo. (2021) Boston Consulting

Group. “Data Alchemy Can Give Decision Making the Golden Touch”

https://www.bcg.com/fr fr/publications/2021/value-of-data-alchemy-decision-making

F. Candelon, M. Gombeaud, B. Ding, Fortune, BCG Henderson Institute (pending

publication May 2022)

Florida R (2002) The rise of the Creative Class. And How it’s transforming work,

leisure, community and everyday life. New Yrok : Basic Books.

Ghebali-Boukhris, L. (2018). Analyse de la contribution d’un nouvel usage des réseaux

sociaux numériques à la connaissance organisationnelle: la curation de contenu

(Doctoral dissertation, Paris, CNAM).

Johnson, D. (2001). What is innovation and entrepreneurship? Lessons for larger

organisations. Industrial and commercial training.

Johnson, S. (2010). Where good ideas come from: The natural history of

innovation. EPenguin.

Julius Colwyn, Rift (2019) MA Arts and Science, Central Saint Martins, ULA,

https://www.arts.ac.uk/__data/assets/pdf_file/0029/191774/MAAS_Class-2019_lowres.p

df

Kauffman, S., & Kauffman, S. A. (1995). At home in the universe: The search for laws of self

organization and complexity. Oxford University Press, USA.

Kelly, N., & Gero, J. S. (2021). Design thinking and computational thinking: a dual

process model for addressing design problems. Design Science, 7.

Koestler, A. (1964). The act of creation.

Kuczmarski, T. D. (1996). What is innovation? The art of welcoming risk. Journal of

Consumer Marketing.

Kuczmarski, T. D. (2003). What is innovation? And why aren’t companies doing more of it?.

Journal of consumer marketing.

Lester, R. K., & Piore, M. J. (2004). Innovation—The missing dimension. In Innovation—The

Missing Dimension. Harvard University Press.

M. Goncalves & G. Heimeriks (2019) The 'adjacent possible' theory and how it can help

research evaluation

https://www.cwts.nl/blog?article=n-r2w294&title=the-adjacent-possible-theory-and

how-it-can-help-research-evaluation

McKelvey, M. (1997): “Using Evolutionary Theory to Define Systems of Innovation”. In:

Edquist, C. (ed): Systems of Innovation: Technologies, Institutions and Organizations. Pinter

Publishers. London, pp. 200-222.

Milan, S., & Van der Velden, L. (2016). The alternative epistemologies of data activism.

Digital Culture & Society, 2(2), 57-74.

Mohamed, S., Png, M. T., & Isaac, W. (2020). Decolonial AI: Decolonial theory as

sociotechnical foresight in artificial intelligence. Philosophy & Technology, 33(4), 659-684.

Oakley, K., Sperry, B., Pratt, A. C., & Bakhshi, H. (2008). The art of innovation: How fine

arts graduates contribute to innovation. Nesta.

Oxman, N. (2016). Age of entanglement. Journal of Design and Science. Brey, P.

(2000). Disclosive computer ethics. ACM Sigcas Computers and Society, 30(4), 10-16.

P., Nenonen, S., Payne, A., & Storbacka, K. (2015). Managing co-creation design: A

strategic approach to innovation. British Journal of Management, 26(3), 463-483.

Pink DH (2006) A Whole New mind : Why right-brainers will rule the future. London :

Cyan Books

Pohlmann, M., Gebhardt, C., & Etzkowitz, H. (2005). The development of innovation

systems and the art of innovation management—strategy, control and the culture of

innovation. Technology Analysis and Strategic Management, 17(1), 1-7.

R. Anadol, (2020) Interview with Sophia Davies for Boston Consulting Group “The

Bionic Renaissance” https://youtu.be/eGe7XqjlDTA

Reynolds, L., & McDonell, K. (2021, May). Prompt programming for large language

models: Beyond the few-shot paradigm. In Extended Abstracts of the 2021 CHI

Conference on Human Factors in Computing Systems (pp. 1-7).

S. Chung (2021) Interview with Sophia Davies for Boston Consulting Group

Ignite, https://youtu.be/sfw_4c1k9bw

Soete, L. (2013). Is innovation always good. Innovation Studies: Evolution and Future

Challenges, 134.

Slavin, K. (2016). Design as participation. Journal of Design and Science.

Tschimmel, K. (2012). Design Thinking as an effective Toolkit for Innovation. In ISPIM

Conference Proceedings (p. 1). The International Society for Professional Innovation

Management (ISPIM).

Van den Broeck, H., Cools, E., & Maenhout, T. (2008) A Case Study of Arteconomy–Building

a bridge between art and enterprise: Belgian businesses stimulate creativity and innovation

through art. Journal of Management