Philosophy Compass

- ARTICLE OPEN ACCESS

Artificial Intelligence and Creativity

Caterina Moruzzi

Institute for Design Informatics, School of Design, The University of Edinburgh, Edinburgh, UK

Correspondence: Caterina Moruzzi (cmoruzzi@[Link])

Received: 30 January 2024 | Revised: 30 January 2025 | Accepted: 16 February 2025

Funding: This work was supported by the Arts and Humanities Research Council (Grants AH/X007146/1 and AH/Y00115X/1) and
Engineering and Physical Sciences Research Council (Grant EP/T022485/1).

Keywords: artificial intelligence | arts | creativity | generative AI | scientific discovery | text‐to‐image | value

ABSTRACT
The question of whether machines can be creative has been at the centre of debates among scholars and
practitioners well before the inception of artificial intelligence (AI) as a recognised field of research. This
paper reviews how some of the key thinkers in the fields of creativity and AI have approached this
question, contextualising their views within the ebbs and fiows of AI technological developments, from
the 1950s until now. The thread of this overview is Margaret Boden's identification of novelty,
surprisingness and value, as the three cardinal features of creativity. This review will retrace the steps of
the quest of artificial intelligence researchers as they strive to replicate each of these three properties
within human‐made machines. The paper closes with a refiection on how the third of these properties,
value, prompts us to consider societal challenges raised by the widespread adoption of AI for creativity
that transcend the question: ‘Can AI be creative?’.

1 | AI's Foundational Quest for Creative develop AI systems that display each of the three
Agents features of creativity identified by Boden: from
novelty, through surprise and finally to value.
In one of its most notable definitions, ‘Creativity is
the ability to come up with ideas or artefacts that ‘Can machines be creative?’. This is a question many
are new, surprising and valuable’. (Boden 2004, 1) thinkers grappled with, way before the origin of AI as
This definition was given by Margaret Boden, a a field of research. Numerous scholars quote the
professor in cognitive science who pioneered the words written by Ada Lovelace, an English
field of philosophy of cognitive science. Boden is also mathematician who collaborated with Charles Bab-
credited with having widened the investigation of bage on the first prototype of a digital computer, as
creativity beyond its traditional focus on human and one of the first objections against the possibility for
animal creativity to encompass computational machines to be creative (Boden 2004; Bown 2021;
systems. Her definition of creativity was destined to Kind 2022; Natale and Henrickson 2022). In
become the reference against which forthcoming particular, Lovelace's claim against machine
studies on creativity would have confronted creativity refers to the inability of the computational
themselves with. This paper will follow in the same systems, which she and Babbage were developing to
footsteps, starting from Boden's definition to retrace display the first of the three features identified by
the history of the relationship between creativity and Boden as essential to creativity, that is, novelty:
artificial intelligence (AI). In particular, it will
review how researchers have strived to

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided
the original work is properly cited.
© 2025 The Author(s). Philosophy Compass published by John Wiley & Sons Ltd.

Philosophy Compass, 2025; 20:e70030 1 of

8
[Link]
 The Analytical Engine has no pretensions the concept underpinning the so‐called ‘AI‐effect’
whatever to originate anything. It can do (McCorduck 2004), that is the phenomenon by
whatever we know how to order it to which once
perform. It can follow analysis; but it has no
power of anticipating any analytical
relations or truths. Its province is to assist us
in making available what we are already
acquainted with.
(Lovelace 1843, Note
G)

This objection, known as the ‘Lovelace objection’ was

contested more than a century later by another
pioneer of modern computer science: Alan Turing. In
his infiuential paper ‘Computing machinery and
intelligence,’ (1950) Turing points out how
Lovelace's claim was grounded on the evidence that
she had available about the Analytical engine, while
‘It is quite possible that the machines in question had
in a sense got this property’. (Turing 1950, 450)
Lovelace's objection should, therefore, according to
Turing's reading, be interpreted only as applicable to
the capabilities of the pioneering computational
systems which were created in Lovelace's time, and it
should not be extended to their future developments
and innovations. In essence, Turing's response to
Lovelace's objection is based on the fact that we can
conceive of human brains as machines; therefore, if
human brains can produce something new, machines
should also be able to do the same (Turing 1950).1

The term ‘Artificial Intelligence’ was coined in the

same years in which Turing was developing his
theories about human and artificial brains. Its
creation is attributed to John McCarthy, an
American computer scientist who co‐organised a
summer workshop in 1956 held at Dartmouth
College, USA, which conventionally marks the start
of AI as a research field. One of the co‐organisers of
the workshop was Marvin Minsky, American
cognitive and computer scientist who contributed to
many of the first advances in the theory and
applications of AI. A report that Minsky published
the same year of the Dartmouth workshop,
‘Heuristic Aspects of the Artificial Intelligence
Problem,’ (1956) is a testament of how, from its
inception, research in AI aspires to replicate human
capabilities, including creativity.

As evidenced by Turing's foundational paper, the

possibility for machines to create something novel is
already widely accepted in the mid‐1950s. Minsky's
refiections focus rather on the ‘sur-prise’ effect that
creativity originates in the spectators. In Minsky's
words:

It is perfectly true that no machine has been

built which remains very impressive after
careful study. But this may be more a
refiection on the nature of being impressed
than a refiection on the nature of machines or,
for that matter, human brains.
(Minsky 1956, 3)

Surprise is a property that is more difficult to

achieve by machines, but not primarily because of a
lack of capacities of the latter, rather because of the
observers' attitude when judging the process they are
witnessing.2 In this report, Minsky also articulates
2 of Philosophy Compass,
15 2025
advancements are made in AI, particularly in nomenon (Miller 1996). In parallel, the use of AI for
accomplishing tasks perceived as indicators of scientific discovery is usually less controversial
intelligence, such as chess playing, these (Langley 1987), whereas
accomplishments tend to be downgraded as not truly
indicative of intelligence or, in the context of this
discussion, creativity. And one of the reasons for
this, according to Minsky, should be found in the
loss of the surprise effect that the said achievement
suffers, when we are able to understand how it was
possible to reach it:

as soon as any performance has been either

mechanised or ‘explained,’ it is dropped,
with appropriate qualifications, from the
list of intelligent performances. And
certainly, if a process is understood or
‘explained,’ then by that fact one no longer
needs the notion of creativity in his
description, and it seems only fair to drop it
from the list of ‘creative’ acts.
(Minsky 1956, 3)

Animated by the optimism which characterised this

first stage of AI research, early pioneers such as Allen
Newell and Herbert Simon conducted ground‐
breaking experiments to demonstrate that AI can
indeed generate novel concepts (Newell, Shaw, and
Simon 1962; Simon 1985). Many are the theorists
who support the relevance of novelty for creativity
(Beardsley 1965; Boden (2004); Kronfeldner
(2009); Simonton (1999); Stokes (2008, 2011,
among others). However, novelty comes in degrees.
Something might be novel in respect to the whole
history of humankind (what Boden calls ‘historical
creativity’, Boden 2004, 43–45), or it might be novel
just for the individual who is undertaking the
creative process in question (Boden's ‘psychological
creativity’). In the case of Newell and Simon's
experiments, attributing historical creativity would be
incorrect as most of these experiments were aimed
at recreating past achievements.

For example, among their most notable experiments

we find ILLIAC, a programme designed to compose
music using Pal-estrina's rules of counterpoint
(Newell, Shaw, and Simon 1962), BACON, which
rediscovered Kepler's third law and other laws of
physics (Simon 1985), and the Logic Theorist, a
programme ‘capable of discovering proofs for
theorems in elementary symbolic logic’. (Newell,
Shaw, and Simon 1962, 67) Newell and his
collaborators define creativity as ‘a special class of
problem‐solving activity characterised by novelty,
unconventionality, persistence and difficulty in
problem formulation’. (Newell, Shaw, and Simon
1962, 66) Their refiections and experiments fall
under an interpretation of creativity as something
that is understandable, measurable and reproducible.
In their view, we should not fall prey of the illusion
that creativity is something reserved to a few
creative ‘geniuses.’ Ordinary people can be equally
creative, the only advantage that the first have over
the latter is only their better heuristics abilities
(Simon 1985).

Although the interpretation of creativity as problem‐

solving might be acceptable in the field of scientific
discovery, it is not so in the field of the arts, where
the Romantic tradition has historically shrouded
creativity in an aura of awe and mystery that makes
of creativity an unexplainable and enigmatic phe-
3 of
15
the application of AI to instances of artistic creativity The primary factor contributing to the increase in
has been met with more resistance (as an example debates around creativity and AI was the
among many, see Dreyfus 1992). introduction of neural

This is partly due to the fact that the concept of

surprise still remains elusive in AI. Something can be
surprising in the sense that it is not easily
predictable, or because it provokes unexpected
reactions in the audience. In the first sense of
‘surpris-ing’, it could be argued that everything that
AI models do could in principle be predicted (Audi
2018).

One way to simulate unpredictability in machines is

through randomness, which has been exploited as a
device by the pioneers of computer art in the 1950s.
Figures such as Max Bense advocated for generative
aesthetics, leveraging computers and algorithms to
create art that surprises through stochasticity, thus
marking a convergence of the concepts of rule‐
bound creation and chaotic creativity (Nake, Nees,
and Cohen 2019; Noll 1966; Pontus Hultén 1968).
The computer artists of this time, like Georg Nees,
Frieder Nake, and Michael Noll, are information
engineers: the choice of technology and the mastery
of its intricacies are still necessary in order to
control the machines in delivering satisfactory
outputs. In this era, artists navigate a terrain where
rules, freedom, technical skills, and conceptual
ideation intermingle, leading to new dimensions of
artistic expression.

In the early 1970s, Harold Cohen, another

artist/engineer, developed one of the first and most
famous computational programmes for generating
drawings: AARON (Garcia 2016). The collaboration
between Cohen and AARON spanned many decades,
and in the mid‐1990s AARON was developed into a
system that could colour forms and figures, in
addition to drawing them. Despite this, Cohen was
still reluctant to consider AARON creative,
identifying rather the creativity in the exchange
between the human and the machine:

Creativity—this particular example of

creativity—lay in neither the programmer
alone nor in the program alone, but in the
dialog between program and programmer; a
dialog resting upon the special and
peculiarly intimate relationship that had
grown up between us over the years.
(Cohen 2010, 9)

The debates around creativity and AI brought to the

fore by Boden, as well as by artists such as Cohen in
the last decades of the 20th century, dwindled
somehow at the end of the century in
correspondence with the second AI winter, a period
of disillusionment following a rise in expectations for
the potentiality of AI which was not matched by its
achievements. But the field of AI was on the brink of a
revolutionary moment, destined to catalyse a surge in
debates surrounding creativity and AI: the success of
neural networks and deep learning.

2 | The Rise of Debates on Creativity

and AI

4 of Philosophy Compass,
15 2025
networks, a method that aims to imitate how The first group usually appeals to fundamentally
neurons work in animal brains to achieve better human and social motives behind creativity that
performance in machines. The history of artificial cannot be shared by
neural networks dates back to a pre‐AI era.
McCulloch and Pitts theorised the first
computational model of a neuron in 1943 (McCulloch
and Pitts 1943) but it was not until 1957, when
Frank Rosenblatt at the Cornell Aeronautical Lab-
oratory built the first implementation of an artificial
neural network, or perceptron, that its potential for
AI innovation became clear, starting the first AI
summer (Rosenblatt 1958).

A perceptron is inspired by certain aspects of

biological neurons. This ‘artificial neuron’ takes
multiple input signals, each with an associated
weight which represents the importance of that
input. The inputs are then summed up, and if the
total surpasses a certain threshold, the perceptron
‘fires’ or produces an output signal. Deep learning
systems are made of perceptrons, or neural
networks, which are assembled in multiple layers
that progressively learn more complex features and
representations of the training data that are
provided as input. This layered learning helps the
network better understand the underlying patterns
in the data, leading to more accurate predictions.
Essentially, the deeper the network, the more
capable it becomes of handling complex and large‐
scale data by capturing subtle relationships that
simpler models might miss. The capacity that these
models have for open‐ended learning and
generalisation beyond training data has a clear
effect on their possibility of generating novel
output. This novelty can be interpreted not just in
Boden's ‘psychological’ sense but also in an
‘historical’ one. An example of this is when, in the
late 2010s, the company DeepMind achieved a
remarkable success in the domain of biology, with
the AI programme AlphaFold, addressing the
longstanding challenge of deciphering the three‐
dimensional structure of proteins and outperforming
other methods by accurately predicting their
structures (Service 2020).

Whether AI can be creative become a subject of

public debate in the mid 2000s, thanks in particular
to generative adversarial networks (GANs) and
AlphaGo. Introduced in 2014 as a new kind of
unsupervised generative algorithm, GANs generate
high‐quality, realistic data that mirrors the
characteristics of the training set and found a
particularly successful application in the field of
style transfer and image synthesis (Goodfellow
2014). Developed by DeepMind, AlphaGo marked a
pivotal advancement in the domain of board games.
In 2016, around 200 million people watched live as
AlphaGo defeated the world‐master Lee Sedol in a
tournament of the ancient game of Go, reaching
international fame (Halina 2021; Silver 2017).
These achievements drew considerable mediatic
attention to a field that, until then, was mostly of
interest only to researchers and technologists. With
this renewed interest, scholars felt the necessity of
going back to the old question: ‘Can a machine be
creative?’

Currently, the field of research on creativity and AI

broadly splits between those who are sceptical about
the potential for AI to be creative, and those who
instead do not exclude the possibility for AI to
display creative properties.

5 of
15
machines, for example, ‘human intent, inspiration, a authenticity is impossible for AI’ (M. Runco 2023,
desire to express something’. (Hertzmann 2018, 1)
Sean Dorrance Kelly, in an article which became a
manifesto of the sceptical views in respect to the
possibility of AI to be creative, affirms: ‘We may be
able to see a machine's product as great, but if we
know that the output is merely the result of some
arbitrary act or algorithmic formalism, we cannot
accept it as the expression of a vision for human
good. […] For this reason, it seems to me, nothing
but another human being can properly be
understood as a genuinely creative artist’. (Kelly
2019).

Novelty and surprise are not enough to attribute

creativity to AI. Philosophers such as Berys Gaut and
Mark Runco outline essential aspects of creativity,
including intentionality, agency, and authenticity,
contending that the challenge in acknowledging AI
creativity on the same level of human creativity stems
from AI's lack of these attributes (Gaut 2010; M.
Runco 2023). Runco ties intentionality to problem‐
finding (Csikszentmihalyi 1988), but it can be just as
easily linked to the requirement of agency expressed
by Gaut and other authors (Brainard 2023; Kieran
2014; Stokes 2008). Agency has, in fact, frequently
been equated to the capacity of initiating events
with an intention (Malafouris 2008). Autonomy, goal‐
directedness, accountability, reactivity, are only some
of the many criteria that have been discussed in the
literature, and according to which we should
determine whom we should or should not attribute
agency to (Moruzzi 2023; Schlosser 2019).3

Gaut defines creativity as ‘the capacity to produce

original and valuable items by fiair. So creativity […] is
a particular exercise of agency. As such it is open to
agents, whether human or not, that have the requisite
capacities’ (Gaut 2010, 1041). Not every kind of action
performed by an agent can be classified as creative, but
only the ones that exhibit a ‘relevant purpose’ (Gaut
2010, 1040). Lovelace's objection, with which this paper
opened, had agency as an implicit requirement, too.
Indeed, the analytical engine was able to perform only
following the instructions given by whoever was
operating it. The agency of the human operator is,
thus, essential for the machine to generate output.

In a recent article, M. Runco (2023) suggests an

update to the bipartite Standard Definition of
Creativity, which includes originality and
effectiveness—which can be understood as ‘utility,
appropriateness, or fit’ (M. Runco 2023, 1)—as key
features (M. Runco and Jaeger 2012), in order to
distinguish human from artificial creativity. Runco
asserts that existing AI systems might meet the
criteria for creativity under the Standard Definition.
However, doubts persist regarding the potential for AI
to exhibit creativity, particularly in a manner comparable
to that of humans. Intentionality and authenticity are
the two candidates that he selects as an addition to the
Standard Definition of Creativity. The motivations given
by Runco for the addition of authenticity to the
Standard Definition, is threefold: due to (1) its role
for self‐actualisation, understood as a state in which
an individual is true to themselves, expressing their
genuine thoughts and feelings, (2) the necessity to
broaden the applicability of the definition of creativity to
non‐Western cultures, where authenticity is
considered as important as, or possibly even more
important than, originality when it comes to creativity
(M. Runco 2023, 1), and (3) to ‘the fact that
6 of Philosophy Compass,
15 2025
1). This last claim, and Runco's overall proposition to
update the Standard Definition of creativity in order
to account for the un-likelihood for AI to be creative,
might be questioned as an example of the ‘AI effect’
that was mentioned earlier: as soon as researchers
develop AI systems capable of displaying the features
that we attribute to creativity, we promptly discount
these accomplishments as lacking genuine creativity
and alter the criteria for evaluation.

Lastly, even assuming that in future, more powerful,

systems will be able to achieve properties such as
agency, intentionality, and authenticity, some assert
that we might still reject claims in favour of AI
creativity on the basis of the fact that ‘Art is
fundamentally a social interaction, and thus can
only be made by social agents’. (Hertzmann 2018,
20).

Theorists who instead wish to support the possibility

for AI to be creative may take two different routes,
either arguing for a nonanthropocentric definition
of creativity which does not include features that
are paradigmatically and exclusively human
(Moruzzi 2021; Newell, Shaw, and Simon 1962), or
pointing at the impressive achievements in the field
of AI semi‐autonomous generation as a testimony to
the fact that AI should be deemed ‘more than a tool’
(Mazzone and Elgammal 2019, 9). A more nuanced
approach is taken by scholars who, although
arguing that AI systems cannot be considered
creative in themselves on the basis of similar reasons
to the ones raised by the first group—for example,
the lack of intentions and goals— acknowledge the
potential of AI of opening up new opportunities for
human artistic creativity (Anscomb 2022, 16). Even
without envisioning a radical future of human–AI
‘anthrobotic’ hybridisation (Miranda 2020, 597),
which combines the features and abilities of the two,
these views welcome a refiection on how human–AI
interaction in the creative field may enhance human
creativity and innovation.4

Debates around creativity and AI gathered steam very

recently, when, over the summer 2022, a new
technology for artistic creation took the internet by
storm: Generative artificial intelligence (GenAI).
GenAI models have been leveraged by artists in
their practices for much longer (Franco 2022), but
only with large language models, such as OpenAI's
ChatGPT,5 and text‐to‐image systems, such as
Midjourney or Stability AI's Stable Diffusion,6 a wide
and not necessarily computer‐literate audience had
access to this technology (Cetinic and She 2022;
Epstein et al. 2023). The availability, user‐friendly
interface, and speed in producing output of an
impressive level of quality, in comparison to what
was being produced by AI technologies even just a
few months prior, contributed to endowing GenAI
with an aura of magic.

Although it is possible to push‐back against the

argument that these models produce anything that
is ‘new’—as they regurgitate content that has been
previously created (Doshi and Hauser 2024), the
number of both academic research and media
coverage that has focused on them undoubtedly
vouches in favour of the fact that their outcomes
can be considered ‘sur-prising’, insofar as they have
provoked a reaction in the audience. The element of
surprise that AI was lacking, seems now to have been
achieved (Moruzzi 2020; M. Runco 2023).

7 of
15
 the right solutions for a responsible and constructive
3 | Navigating the Future of AI and
interaction between humans and AI in the field of
Creativity artistic creativity, can therefore have rippling

The impact of technology on creative processes is not

a novelty brought about by AI. Both the fields of art
and science have been transformed by technological
innovations.

The invention of the telescope at the turn of 16th

and 17th centuries is traditionally regarded as one of
the most infiuential technological innovations in
science (Helden 1977). It not only enabled the
observation of previously unknown celestial bodies in
the cosmos but also transformed the relation
between the human and the universe.

The music industry is maybe the first creative

industry where the impact of technology had an
international resonance. Recording technology
which started in the late 18th century with
phonograph cylinders, gramophone records and, later
on, electrical microphones, allowed for the
reproduction and distribution of musical
performances, altering the landscape of music
consumption and the revenue models of the music
sector. One of the most apparent consequences was
the gradual replacement of live performances—and
musicians—in a variety of contexts: from movie
theatres to public events. This shift led to the
emergence of recording labels as key players in the
industry. Musicians sought to address the
disproportionate power dynamics with record labels
through various strategies, one of which was the
creation of independent production and distribution
channels to circumvent the traditional oversized
control of record labels on artist promotion,
production and sales. This gradual transformation
was conducted in parallel to many other sectors of the
cultural economy which were seeking more fiexible
and independent work arrangements—the so‐called
gig‐economy model (Cloonan and Williamson 2023).7

The relationship between creativity and

technological innovations forms a virtuous cycle,
where the advancement of one element propels and
stimulates the development of the other. This is
evident, among many other examples, in the
development of piano technology. Piano makers in
the 18th and 19th century expanded the keyboard's
range and the durability and stability of the
instrument's frame to accommodate the creative
demands of musicians such as Beethoven and Liszt
who asked for a broader range of expression
(Giordano 2016). Both fields benefited from the
exchange: the technology of the instrument
considerably improved, and musicians achieved the
necessary means to allow for greater dynamic
contrasts and nuanced playing.

GenAI technology is likely to have a similar kind of

disruptive effect on the creative sector as the
innovations just mentioned. At the same time, it has
also the potential of furthering similar beneficial
advancements in both science and the arts.
Throughout history, the arts have consistently been
at the vanguard in driving innovation, serving as a
catalyst for pushing the frontiers of knowledge and
inspiring transformative change across various
domains. With the upsurge of GenAI techniques for
the generation of creative content, artists are
occupying yet again a pioneering position. Finding
8 of Philosophy Compass,
15 2025
effects in many other fields that are affected by the polarisation between the two sides of
transformative impact of AI.

Still, one issue remains to be settled: how should we

respond to the question ‘Can machines be creative?’
This is the question with which this paper opened
and that does not seem to have found an answer,
yet. The sceptics still have one argument up their
sleeves: when attributing the capability to create
novel and surprising material to AI, we are adopting
a product‐first account of creativity, namely, we are
assessing the creativity of an agent or a process on
the basis of the feature of the generated product (in
this case, novelty and surprisingness) (Currie and
Turner 2023; Gaut and Kieran 2018).

From a product‐first perspective, arguing that AI can

be creative is easier: AI can create novel output–for
example, AlphaFold predicts structures of proteins
that were previously unknown or not solved
experimentally, ChatGPT generates strings of text
that are contextually driven and not a mere
repetition of the training data, and text‐to‐image
models can synthesise visual elements in ways that
were not explicitly seen in its training data—and this
output is surprising as it is difficult to predict and it
provokes a reaction in the audience. By arguing this,
however, we are liable to a similar observation as
the one raised by Minsky: ‘no machine has been
built which remains very impressive after careful
study’. (Minsky 1956, 3) When switching from a
product‐first to a process‐first account, other ele-
ments of creativity, such as intentionality, agency
and autonomy, suddenly gain more relevance
(Moruzzi 2021). AI is a tool that facilitates and
enables the creation of something new and
surprising, it is not the main driving force in the
creative process. Hence, attributing creativity to AI
from a process‐first perspective is questionable.

In addition, one of the three features of creativity

identified by Boden remains to be addressed:
‘value.’8 Content generated by and through AI might be
novel and surprising, but is it valuable? If we do not
merely interpret value as endorsement by critics,
galleries, record labels and the general public of the
aesthetic merit of the products generated by AI (Paul
and Stokes 2023), but instead, more broadly, as the
promotion of human well‐being and human fiourishing
(Woodruff 2001), then for AI to achieve this last
property seems a less obvious task.9 Focusing on the
concept of value, interpreted in the second way, also
allows us to distance ourselves from the original
question ‘Can machines be creative?’, and finding an
answer to this question suddenly does not appear as
the most urgent issue that we need to face. The
impact that GenAI is having on different
communities and sectors calls attention to the
necessity of focusing less on theoretical questions,
such as this one, and more on the social
implications of the widespread adoption of GenAI
(Steyerl 2023).

Just as the polarisation between ‘Wonder and Panic’

in response to the latest achievements in AI
technology for the generation of content tends to
obscure more urgent issues, such as the scraping of
content for the training data, the lack of attribution
to the original creators of the content, and the
hidden labour that goes into the development of the
dataset and into the moderation of the generated
content (Goetze 2024; Sarkar 2023),10 so does the

9 of
15
the debate around whether AI can be deemed
‘creativity’ and ‘AI’ is a duo that inspires and
creative or not. Rather, it would be more fruitful to
prompts an enrichment of the human creative
focus on how AI is changing creative processes and
experience.
how we can act accordingly to prevent
misuses of the technology from happening (Rafner et al. 2023).

Within the computational creativity and the human– between humans and AI is necessary for building a
computer interaction community, the mixed‐initiative constructive co‐creativity relationship, we need the
co‐creativity field of study has this among its core combined effort of researchers, artists, professionals
aims. The mixed‐initiative co‐creativity paradigm and enthusiasts to ensure that
was first presented in a paper in 2014 by
Yannakakis, Liapis, and Alexopoulos (2014). It
explores how human and computational agents can
contribute to the generation of creative content by
acting in synergy, examining the roles and
interaction patterns between human users and
computational systems in co‐creative processes
(Davis 2021; Deterding et al. 2017). This paradigm
stands in the middle of a spectrum, between more
traditional creativity support tools, such as
computer‐aided design (CAD), where the agency is
in the hands of the human user, and computational
creativity, where the artificial agent generally has
more control over the process in an autonomous or
semi‐autonomous way. The increasingly widespread
use of GenAI models is shifting the agency toward
the artificial agent's side of the spectrum.

This shift may be responsible for a change in the

views of thinkers who argue that AI cannot be
creative on the basis of a lack of agency, where
agency is understood as the capacity of autonomous
control over its own actions. In the creative scenario,
transformed by the widespread adoption of AI tools,
it is urgent to investigate how the different balances
of agency and control between humans and
machines can change the user experience and
perception in creative processes that are performed
together with technology (Evans 2023). The call for
an increased attention to the human–machine
interaction, goes in the same direction as the
refiections made by Anscomb (2022) and other
researchers on the investigation of what it is that
underpins the support that AI can provide to human
creativity.

Agency attribution, in one or more of the

interpretations of the term mentioned in Section 2, is
essential also for the ascription of causal and legal
responsibility, as it allows agents to distinguish
events that they caused from those for which they are
not responsible for. In the creative domain, this tightly
relates to the attribution of intellectual property.
Recent legal disputes over copyright claims over
content generated with and by AI, including artists
suing AI companies, and the Hollywood writers'
strike advocating for guardrails against the use of AI
by production studios, underscore critical
apprehensions shared by creative communities about
lost jobs and stolen labour in the digital art
landscape.11

In this scenario, philosophers are called to answer

questions that are maybe narrower, but more
immediately socially relevant than the question ‘Can
machines be creative?’ And, in doing so, they
necessarily need to widen their debates to include
experts from other disciplines, from law, to
computer science, psychology, design and the arts.
Just as a synergy between creativity and
technological innovation is key to prompt a virtuous
cycle which benefits both fields, and the alignment
10 Philosophy Compass,
of 15 2025
Conflicts of Interest
The author declares no confiicts of interest.

Endnotes
1
This thought has been more recently echoed by the Google
DeepMind principal research scientist Murray Shanahan
(Shanahan 2020).
2
In his 1950 paper, Turing affirmed that machines already
took him by surprise ‘with great frequency’ (Turing
1950, 450), although he acknowledges, as Minsky does,
that the surprise might be due to ‘some creative mental
act’ (ibid., p. 451) on his part, rather than to the
machine's capabilities.
3
Some key definition of agency that have been given in
the philosophical literature can be found in Anscombe
1971; Bratman 2007; Davidson 1963; Kane 2011; Latour
2007.
4
For an overview on debates around creativity, see the
Stanford Encyclopaedia of Philosophy entry by Paul and
Stokes (2023).
5
OpenAI's blog post available at
[Link]
6
Available at [Link]
callbackUrl=%2Fex plore and [Link]
7
And, indeed, the term ‘gig’ in gig‐economy originally had its
roots in the
music industry, referring to individual musicians' live
performances.
8
Value is discussed as a condition for creativity in
Amabile (1996), Boden (2004), Carruthers (2011),
Kieran (2014), and Novitz (1999), among others.
9
It should be noted that value is a contested requirement
of creativity which scholars do not unanimously agree on.
For a discussion on this, see Kind 2022, 25–27.
10
See Alan Warbuton's ‘The Wizard of AI’, a 20 minute,
99% AI‐generated visual essay developed by Warbuton
for the Open Data Institute's 2023 summit. Video
available at:
[Link]
11
The copyright and legal disputes are covered here:
[Link]
[Link]/commentisfree/2023/aug/26/ai‐generated‐art‐
copyright‐law‐recent‐entrance‐paradise‐creativity‐machine
and here:
[Link]
ai‐art‐copyright‐legal‐lawsuit‐s table‐diffusion‐midjourney‐
deviantart. Information about the Hollywood writers' strike
available at: [Link]
ke‐ai‐provisions‐precedents.

References
Amabile, T. M. 1996. Creativity in Context: Update to the
Social Psychology of Creativity. Boulder, CO: Westview.
Anscomb, C. 2022. “Creating Art With AI. Odradek.”
Studies in Philosophy of Literature, Aesthetics, and New
Media Theories 8, no. 1: 13–51.
Anscombe, G. E. M. 1971. Causality and Determination.
Cambridge: Cambridge University Press.
Audi, R. 2018. “Creativity, Imagination, and Intellectual
Virtue.” In Creativity, Imagination, and Intellectual Virtue,
edited by B. Gaut and M. Kieran, 24–41. New York:
Routledge.
Beardsley, M. C. 1965. “On the Creation of Art.” Journal of
Aesthetics and Art Criticism 23, no. 3: 291–304.
[Link] jaac23.3.0291.
Boden, M. 2004. The Creative Mind: Myths and
Mechanisms. 2nd ed. London: Routledge.

11
of 15
Bown, O. 2021. Beyond the Creative Species: Making Gaut, B., and M. Kieran. 2018. “Philosophizing About
Machines that Make Art and Music. Cambridge (MA): MIT Creativity.” In Creativity and Philosophy, edited by M.
Press. Kieran and B. Gaut, 1–22. London: Routledge.
Brainard, L. 2023. “The Curious Case of Uncurious
Creation.” Inquiry: 1–31.
[Link]
Bratman, M. E. 2007. Structures of Agency: Essays.
Oxford: Oxford University Press.
Carruthers, P. 2011. “Creative Action in Mind.”
Philosophical Psychology
24, no. 4: 437–461.
[Link]
Cetinic, E., and J. She. 2022. “Understanding and Creating
Art With AI: Review and Outlook. ACM Transactions on
Multimedia Computing.” Communications, and
Applications (TOMM) 18, no. 2: 1–22. [Link]
org/10.1145/3475799.
Cloonan, M., and J. Williamson. 2023. “Musicians as
Workers and the Gig Economy.” Popular Music and Society
46, no. 4: 354–370. [Link]
org/10.1080/03007766.2023.2231266.
Cohen, H. 2010. “Driving the Creative Machine.” Talk
given at Orcas Center, Crossroads Lecture Series.
[Link].
Csikszentmihalyi, M. 1988. “Solving a Problem Is Not
Finding a New One: A Reply to Herbert Simon.” Journal
New Ideas in Psychology 6, no. 2: 183–186.
[Link]
Currie, A., and D. Turner. 2023. “Creativity Without
Agency: Evolutionary Flair & Aesthetic Engagement.” Ergo
an Open Access Journal of Philosophy 10, no. 0: 4.
[Link]
Davidson, D. 1963. “Actions, Reasons, and Causes.” Journal
of Philosophy 60, no. 23: 685–700.
[Link]
Davis, N. 2021. “Human‐Computer Co‐creativity: Blending
Human and Computational Creativity.” Proceedings of the
AAAI Conference on Artificial Intelligence and Interactive
Digital Entertainment 9, no. 6: 9–12.
[Link]
de Miranda, L. 2020. “Artificial Intelligence and
Philosophical Creativity: From Analytics to Crealectics.”
Human Affairs 30, no. 4: 597–607.
[Link]
Deterding, S., J. Hook, R. Fiebrink, et al. 2017. “Mixed‐
initiative Creative Interfaces.” In CHI 2017 Extended
Abstracts ‐ Proceedings of the 2017 ACM SIGCHI
Conference on Human Factors in Computing Systems:
Explore, Innovate, Inspire, 628–635. Association for
Computing Machinery.
[Link]
Doshi, A. R., and O. P. Hauser. 2024. “Generative AI
Enhances Individual Creativity but Reduces the Collective
Diversity of Novel Con-tent.” Science Advances 10: 28.
[Link]
Dreyfus, H. 1992. What Computers Still Can't Do: A
Critique of Artificial Reason. Cambridge (MA): MIT Press.
Epstein, Z., A. Hertzmann, M. Akten, et al. 2023. “Art and
the Science of Generative AI.” Science 380, no. 6650:
1110–1111. [Link] 1126/science.adh445.
Evans, K. D., S. A. Robbins, and J. J. Bryson. 2023. “Do We
Collaborate With What We Design?” Topics in Cognitive
Science 00: 1–20. [Link] org/10.1111/tops.12682.
Franco, F. 2022. The Algorithmic Dimension: Five Artists in
Conversation. Cham: Springer.
[Link]
Garcia, C. 2016. “Harold Cohen and AARON – A 40‐ Year
Collabora-tion.” [Link]
cohen‐and‐aaron‐a‐40‐year‐collaboration/.
Gaut, B. 2010. “The Philosophy of Creativity.” Philosophy
Compass 5, no. 12: 1034–1046.
[Link]
12 Philosophy Compass,
of 15 2025
Giordano, N. 2016. “The Invention and Evolution of the Moruzzi, C. 2021. “Measuring Creativity: An Account of
Piano.” Natural and Artificial Creativity.” European Journal for
Philosophy of Science 11, no. 1: 1–20.
Acoustics Today 12, no. 1: 12–19. [Link]
Goetze, T. S. 2024. “AI Art Is Theft: Labour, Extraction, and
Exploitation, or, on the Dangers of Stochastic Pollocks.”
arXiv:2401: 06178.
Goodfellow, I., J. Pouget‐Abadie, M. Mirza, et al. 2014.
“Generative Adversarial Nets.” In Proceedings of the 27th
International Conference on Neural Information
Processing Systems, Vol. 2, 2672–2680.
Halina, M. 2021. “Insightful Artificial Intelligence.” Mind &
Language
36, no. 2: 315–329. [Link]
Helden, A. 1977. “The Invention of the Telescope.”
Transactions of the American Philosophical Society 67,
no. 4: 1–67. [Link]
1006276.
Hertzmann, A. 2018. “Can Computers Create Art?.” Arts 7,
no. 2: 18. [Link]
Kane, R., ed. 2011. The Oxford Handbook of Free Will. 2nd
ed. Oxford: Oxford University Press.
Kelly, S. D. 2019. “A Philosopher Argues That an AI Can’t
Be an Artist.”
[Link]
argues‐that‐an‐ai‐can‐never‐be‐an‐artist/.
Kieran, M. 2014. “Creativity as a Virtue of Character.” In
The Philosophy of Creativity: New Essays, edited by E. S.
Paul and S. B. Kaufman, Vol. 2014, 125–144. Oxford/New
York: Oxford University Press. [Link]
org/10.1093/acprof:oso/9780199836963.003.0007.
Kind, A. 2022. Imagination and Creative Thinking.
Cambridge: Cambridge University Press.
Kronfeldner, M. 2009. “Creativity Naturalized.”
Philosophical Quarterly
59, no. 237: 577–592. [Link]
9213.2009.637.x.
Langley, P. 1987. Scientific Discovery: Computational
Explorations of the Creative Processes. Cambridge (MA):
MIT press.
Latour, B. 2007. Reassembling the Social: An
Introduction to Actor‐Network‐Theory. Oxford: Oxford
University Press.
Lovelace, A. A. 1843. “Translation of, and Notes to, Luigi F.
Menabrea’s Sketch of the Analytical Engine Invented by
Charles Babbage.” In Scientific Memoirs, edited by R.
Taylor, Vol. 3, 691–731. London: Richard and John E.
Taylor.
Malafouris, L. 2008. At the Potter’s Wheel: An Argument
for Material Agency. 19–36. Springer.
[Link]
Mazzone, M., and A. Elgammal. 2019. “Art, Creativity, and
the Potential of Artificial Intelligence.” Arts 8, no. 1: 26.
[Link] arts8010026.
McCorduck, P. 2004. Machines Who Think. 2nd ed. Natick,
MA: A. K. Peters, Ltd.
McCulloch, W. S., and W. Pitts. 1943. “A Logical Calculus
of the Ideas Immanent in Nervous Activity.” Bulletin of
Mathematical Biophysics 5, no. 4: 115–133.
[Link]
Miller, A. 1996. Insights of Genius: Imagery and Creativity
in Science and Art. New York: Copernicus.
Minsky, M. L. 1956. “Heuristic Aspects of the Artificial
Intelligence Problem.” In Lincoln Lab., M.I.T., Lexington,
Massachussets, Group Report, 34–55: ASTIA Doc. No.
236885 (M.I.T. Hayden Library No. H‐58).
Moruzzi, C. 2020. “Should Human Artists Fear AI?: A
Report on the Perception of Creative AI.” Proceedings of
the Eighth Conference on Computation, Communication,
Aesthetics & X: 170–185.

13
of 15
Moruzzi, C. 2023. “Creative Agents: Rethinking Agency and Press.
Creativity in Human and Artificial Systems.” Journal of
Aesthetics and Phenomenology 9, no. 2: 245–268. Steyerl, H. 2023. Mean Images, NLR 140/141.
[Link] [Link] issues/ii140/articles/hito‐steyerl‐
mean‐images.
Nake, F., G. Nees, and H. Cohen. 2019. “Re:tracing the
Origins of Digital Media.” In Digital Art through the
Looking Glass. New Strategies for Archiving, Collecting
and Preserving in Digital Humanities, edited by O. Grau, J.
Hoth, and E. Wandl‐Vogt, 27–48. Krems a.d. Donau: Edition
Donau‐Universität.
Natale, S., and L. Henrickson. 2022. “The Lovelace Effect:
Perceptions of Creativity in Machines.” New Media &
Society 26, no. 4: 1909–1926.
[Link]
Newell, A., C. Shaw, and H. Simon. 1962. “The Process of
Creative Thinking.” In Contemporary Approaches to
Creative Thinking, edited by
H. Gruber, G. Terrel, and M. Wertheimer, 63–119. New
York: Atherton Press.
Noll, M. 1966. “Human or Machine: A Subjective
Comparison of Piet Mondrian's 'Composition With Lines'
(1917) and a Computer‐Generated Picture.” Psychological
Record 16: 1–10. [Link]
635.
Novitz, D. 1999. “Creativity and Constraint.” Australasian
Journal of Philosophy 77, no. 1: 67–82.
[Link]
Paul, E. S., and D. Stokes. 2023. “Creativity.” In The
Stanford Encyclopedia of Philosophy, edited by E. N. Zalta
and U. Nodelman, https://
[Link]/archives/spr2023/entries/creativity/.
Pontus Hultén, K. G. 1968. The Machine as Seen at the End
of the Mechanical Age. New York: MoMa.
Rafner, J., R. E. Beaty, J. C. Kaufman, T. Lubart, and J.
Sherson. 2023. “Creativity in the Age of Generative AI.”
Nature Human Behaviour 7, no. 11: 1836–1838.
[Link]
Rosenblatt, F. 1958. “The Perceptron: A Probabilistic
Model for Information Storage and Organization in the
Brain.” Psychological Review 65, no. 6: 386–408.
[Link]
Runco, M. 2023. “AI Can Only Produce Artificial Creativity.”
Journal of Creativity 33: 3.
[Link]
Runco, M. A., and J. Jaeger. 2012. “The Standard Definition
of Crea-tivity.” Creativity Research Journal 24, no. 1: 92–
96. [Link] 1080/10400419.2012.650092.
Sarkar, A. 2023. “Enough With ‘Human‐AI Collaboration’.”
In Extended Abstracts of the 2023 CHI Conference on
Human Factors in Computing Systems, 1–8.
Schlosser, M. 2019. “Agency.” In The Stanford
Encyclopedia of Philosophy, edited by E. N. Zalta, Stanford
University. [Link]
edu/archives/win2019/entries/agency/.
Service, R. F. 2020. “The Game Has Changed. AI Triumphs
at Protein Folding.” Science (New York, N.Y.) 370, no.
6521: 1144–1145. [Link]
org/10.1126/science.370.6521.1144.
Shanahan, M. 2020. “Are Brains Computers?.” Medum.
[Link] com/@[Link]/are‐brains‐computers‐
6cf906b02ca0.
Silver, D., J. Schrittwieser, K. Simonyan, et al. 2017.
“Mastering the Game of Go Without Human Knowledge.”
Nature 550, no. 7676: 354–
359. [Link]
Simon, H. 1985. “What We Know About the Creative
Process.” In
Frontiers in Creative and Innovative Management, edited
by R. Kuhn, 3–
22. Cambridge: Ballinger.
Simonton, D. K. 1999. Origins of Genius: Darwinian
Perspectives on Creativity. New York: Oxford University
14 Philosophy Compass,
of 15 2025
Stokes, D. R. 2008. “A Metaphysics of Creativity.” In New
Waves in Aesthetics, edited by K. Stock and K. Thomson‐
Jones, 105–124. New York: Palgrave‐Macmillan.
Stokes, D. R. 2011. “Minimally Creative Thought.”
Metaphilosophy 42, no. 5: 658–681.
[Link]
Turing, A. M. 1950. “Computing Machinery and
Intelligence.” Mind 59, no. 236: 433–460.
[Link]
Woodruff, D. M. 2001. “A Virtue Theory of Aesthetics.”
Journal of Aesthetic Education 35, no. 3: 23–36.
[Link]
Yannakakis, G. N., A. Liapis, and C. Alexopoulos. 2014.
“Mixed‐initiative Co‐creativity.” In International
Conference on Foundations of Digital Games.
[Link]

15
of 15