See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/372028377

The Rise of CreAltives: Using AI to enable and speed up the creative process

Article · July 2023

DOI: 10.69554/WLDX9074

CITATIONS READS

12 668

1 author:

Andrew Pearson
Intelligencia
8 PUBLICATIONS 49 CITATIONS

SEE PROFILE

All content following this page was uploaded by Andrew Pearson on 03 July 2023.

The user has requested enhancement of the downloaded file.

 The rise of CreAltives: Using AI to
enable and speed up the creative
process
Received: 15th December, 2022

Andrew Pearson
Managing Director, Intelligencia, China

Andrew Pearson is the President of Intelligencia Limited, a leading implementer of analytics, artificial
intelligence (AI), business intelligence, customer experience, data integration, digital marketing and social
media solutions for clients worldwide, including some of the biggest casinos and sportsbooks in the
world. He has written several books on analytics and AI, including The AI Casino and The AI Marketer, as
well as several others in the AI series. Andrew has written articles for several prestigious publications. He
is also a sought-after speaker and lectures all over the world on such topics as AI, machine learning (ML),
casino marketing, analytics and social media. In summer 2023, his first novel, The Dead Chip Syndicate,
will be traditionally published by Brother Mockingbird press. It is the first in his Exotics Trilogy, which is set
in Asia and focused on the world of tech and gambling.

Intelligencia, Hong Kong/Macau, China

E-mail: [email protected]

Abstract The ancient Greeks invented the concept of the muse goddess to be a vessel
that would enter a human’s life and spark long-desired creativity. The Greek gods did not
want to give humans credit for coming up with creativity; however, humans may have
invented something as good as a muse — artificial intelligence (AI). Today, AI, machine
learning (ML), deep learning (DL) and natural language processing (NLP) could be taking
up that ‘muse role’ as well as becoming the workhorse for many a visual artist’s grunge
work, and possibly even their creative work as well. In this paper, we will look at how
modern creative AI technologies can be viewed through the lens of five groups. We will
also explore how recent advances in creative AI offer users the ability to create images,
compose music, animation and even video in ways never before possible — and then
wrap up with final takeaways on the future of artistic creativity in the era of AI.

KEYWORDS: artificial intelligence, personalisation, machine learning, deep learning,

natural language processing, text-to-image, generative AI, customer relationship
management, behavioural marketing, personalisation, creativity, visual art, music,
animation, video, video gaming, images

CREATIVITY we may have found a new muse in artificial

The ancient Greeks believed creativity came intelligence (AI). In their article, ‘Artificial
from the gods, even inventing the concept intelligence in the creative industries: A
of the muse, a goddess who enters an artist’s review’,1 Anantrasirichai and Bull contend
life and sparks creativity. Many a blocked AI has increasingly
writer, painter, musician, filmmaker and
artist has longed for a visit by the mythical ‘been associated with human creativity and
muse to set their imagination alight. Today, artistic practice. As it has demonstrated

© Henry Stewart Publications 2633-562X (2023) Vol. 2, 2 1–14 Journal of AI, Robotics & Workplace Automation   1

Pearson.indd 1 30/01/2023 14:56

 Pearson

abilities to “see”, “hear”, “speak”, landscape, creative projects almost always

“move”, and “write”, it has been applied include many phases of complex iteration,
in domains and applications including: adaptation and process management that
audio, image and video analysis, gaming, are usually cited as the most tedious part
journalism, script writing, filmmaking, of the creative work. Whether managing
social media analysis and marketing.’2 image, video or text files, which can contain
dozens of layers, or coordinating the transfer
Anantrasirichai and Bull believe the of innumerable assets from one system to
maximum benefit from AI can be derived another or to and from the cloud, the flow
when it augments, rather than replaces, from creative intent to the final product is
human creativity.3 They argue creativity a complex and challenging one — one that
demands a high level of innovation and could do with more optimisation.9
routine behaviour is almost antithetical to the The iterations in marketing content
creative process.4 ‘While AI accomplishments that are coming in the next few years will
rely heavily on conformity of data, creativity make the complexity of today’s advertising
often exploits the human imagination to drive landscape look like child’s play. Admittedly,
original ideas which may not follow general some creatives enjoy these repetitive tasks,
rules’, contend Anantrasirichai and Bull.5 as Pfeiffer Reports discovered.10 For some
Albert Einstein would agree, as he famously creatives, working through the repetitive
once said, ‘Creativity is seeing what others see steps needed to execute a creative project
and thinking what no one else ever thought’. is seen as useful and necessary.11 As one
Creatives can build upon their experience, designer put it, ‘Not everything should be
which enables them to think ‘outside the box’ made too easy. The trial-and-error process
while asking ‘What if ’ questions not easily linked to technology is actually important
addressed by constrained learning systems, say for the creative process. Errors are important,
Anantrasirichai and Bull.6 don’t take them away.’12 There is, however, a
For Anantrasirichai and Bull, creative large portion of creatives who would rather
AI technologies can be broken down into spend their valuable time on the endeavours
the following five groups: (1) content requiring inspiration rather than perspiration.
creation; (2) information analysis; (3) ‘Art is the elimination of the unnecessary’,
content enhancement and post-production Pablo Picasso once claimed. This is an
workflows; (4) information extraction and idea that is much more relevant in today’s
enhancement; and (5) data compression (see complicated marketing world than it was
Table 1).7 in the simpler time of Picasso. It is a quote
AI can now read, understand and translate many creatives would agree with, especially
text on smart devices and computers. Virtual when it comes to the most tedious parts
assistant technology can listen to voices, of the creative process. AI could be at the
respond to questions as well as trigger events, forefront of eliminating those unnecessary
such as playing music, ordering food, or tasks: the monotonous, time-consuming
buying products and services online. AI can manual operations most creatives would
view images and recognise objects in them; gladly hand off to someone else, even if that
however, there is no creative thought in any someone else was an AI bot. AI is not just
of these endeavours. for eliminating tasks, however; it could offer
‘There is a clear expectancy that a little creativity of its own.
technology should be there to serve the In their paper, ‘Artificial Intelligence
creative’, states Pfeiffer Reports in its in music and performance: A subjective,
‘Creativity and technology in the age of AI’.8 art-research inquiry’,13 Caramiaux and
Given the complexity of the modern media Donnarumma argue artists are usually the

2   Journal of AI, Robotics & Workplace Automation Vol. 2, 2 1–14 © Henry Stewart Publications 2633-562X (2023)

Pearson.indd 2 30/01/2023 14:56

 The rise of CreAltives: Using AI to enable and speed up the creative process

Table 1: Creative AI technologies

Category Sub-category
Creation Content generation for text, audio, video and game
Animation
Augmented reality (AR)/virtual reality (VR)
Deepfakes
Content and captions
Information analysis Text categorisation
Ads/film analysis
Content retrieval
Fake detection
Recommendation
Content Contrast
Enhancement and post-production Colourisation
Super-resolution
Deblurring
Denoising
Dehazing
Turbulence removal
Inpainting
Visual effects (VFX)
Information extraction Segmentation
Recognition
Tracking
Suspension of disbelief (SOD)
Fusion
3D reconstruction
Data compression

early ones to adopt technology, often using repetitive tasks of the work, with 89 per cent
it in ways that disrupt society. AI, machine of respondents expressing interest in such a
learning (ML) and deep learning (DL) will system (see Figure 1).16 Pfeiffer Consulting
be no exception.14 On the one hand, many reports:17
artists already use ‘ML to enrich the way they
work with their preferred media. On the ‘For creative professionals working
other, some artists use ML, and in particular, frequently with stock images, or other
AI, to shed light onto certain facets of materials such as fonts found on the
these same tools which can be invisible web, AI and ML-based help with these
or taken for granted by public opinion, tasks would be very welcome: 81% of
media, or institutions’, claim Caramiaux and respondents indicated strong interest for
Donnarumma.15 such an assistant.’
When asked which type of assistant
creatives would prefer, the clear favourite was Teaching assistants that could explore
an assistant that reduced drudgery and the new features in the software also received

© Henry Stewart Publications 2633-562X (2023) Vol. 2, 2 1–14 Journal of AI, Robotics & Workplace Automation   3

Pearson.indd 3 30/01/2023 14:56

 Pearson

Figure 1: Which AI- or ML-based assistant system would creatives be interested in?18

strong approval ratings;19 however, assistants motion graphics creatives were also interested
providing creative variations based on a in testing when a viewer’s level of attention
project at hand only received 51 per cent dropped while watching a clip.24 User
approval. Many interviewees expressed interest experience (UX) designers also welcomed
in trying such an assistant, but they did not testing the logic and intuitiveness of a user
expect to rely on it, since they preferred to interface design.25
stay in control of their creative process.20 This
sentiment is certainly a justified one.
One creative assistant suggestion that drew TRANSCENDENT BEAUTY AS A
the most interesting comments was a system SERVICE
that would help anticipate different types In ‘The Art in the Artificial: AI and the
of audience reactions to a given project, creative industries’,26 Davies et al. see AI
based on ML.21 ‘While some respondents within the creative industries breaking down
(almost 18%) had spontaneous strong into three categories: AI involving media/
negative reactions to this type of assistant, domains relevant to creative industries;
over 59% stated strong or extremely strong AI research directly applied to a creative
approval’, according to Pfeiffer Consulting.22 industry-related activity; creative/generative
The type of information the respondents content, such as generative adversarial
wanted to receive using such an assistant networks (GANs), which imitate artists and
was particularly interesting, with several artistic styles. The latter, like DALL-E and
designers mentioning they would like to test MidJourney, has already produced some
the uniqueness of a design.23 Some video and impressive and compelling work.

4   Journal of AI, Robotics & Workplace Automation Vol. 2, 2 1–14 © Henry Stewart Publications 2633-562X (2023)

Pearson.indd 4 30/01/2023 14:56

 The rise of CreAltives: Using AI to enable and speed up the creative process

The first category relates to images, audio must focus ‘less on the details of the actual
and text: outcome of a creative project, and much
more on the multi-layered engagement
‘The raw material of the creative processes necessary to reach and touch’ an
industries: language, imagery and sound audience, argues Pfeiffer Consulting.32 They
can all be digitized and converted into contend:
data. As machine learning algorithms are
often applied to these, AI research of this ‘Understanding the client’s needs and
nature could have future implications for dreams, and putting themselves in the
the creative industries, although it may not shoes of the client’s target audience is
necessarily be directly analysing creative essential for creative professionals. This is
content.’27 what shapes their creative vision, and helps
find inspiration. But the creative project
The second category uses AI to animate is also shaped by technology: Technology
a character in a computer game or, like defines what it is possible to create — and
in Adobe’s Character Animator software, also what it is possible to deliver to the
generates a visual effect in an image, a video audience.’33
or even on a livestream. The third category,
creating generative content as well as the Today’s creative professionals must balance
ability to mimic an artist’s style, has grown their creative vision within the given
rapidly in the past few years. technological framework they are working
Style transfer is a process allowing a style with, while juggling the increasingly
from one medium to be transferred to complex demands of multi-format, multi-
another.28 For example, on AI text-to-image device content delivery, which can both
tools like DALL-E and MidJourney, users enable and constrain an artist.34
add prompts like ‘abstract expressionist’ or
‘in the vein of Warhol’ to utilise a famous
or even not-so-famous artist’s style to their IMAGES AND AI
work. As noted by Davies et al: In early 2021, OpenAI unveiled its picture-
making neural network, DALL-E, with
‘An example of an artistic application a string of surreal and cartoonish images
is Gene Kogan’s Cubist Mirror where produced on demand that would probably
a viewer standing in front of a screen have made the service’s namesake artist,
finds their recorded image alternatively Salvador Dali, immensely proud. As Will
transformed into the styles of different Douglas Heaven explains in his article,
artists such as Klimt, Munch and ‘This horse-riding astronaut is a milestone
Hokusai.’29 on AI’s long road towards understanding’.35
‘DALL-E’s avocado armchairs had the
Caramiaux et al.30 discovered AI in the essential features of both avocados and chairs;
current media and creative industries could its dog-walking daikons in tutus wore the
be found in three areas: creation, production tutus around their waists and held the dogs’
and consumption. Although technology lets leashes in their hands.’36 An AI system getting
creatives realise the potential of their vision, these and other important details right was
it also defines the boundaries of engagement truly cutting-edge.
by providing an ever-increasing number To explore how AI can have an impact
of distribution channels and platforms a on the creation of images with AI, the
creative professional must accommodate, says authors of this paper tested out MidJourney,
Pfeiffer Consulting.31 This means a creative a text-to-image service calling itself ‘An

© Henry Stewart Publications 2633-562X (2023) Vol. 2, 2 1–14 Journal of AI, Robotics & Workplace Automation   5

Pearson.indd 5 30/01/2023 14:56

 Pearson

independent research lab. Exploring new texts. The more training data to pull
mediums of thought. Expanding the from, the better these systems become at
imaginative powers of the human species.’ recognising patterns and making sense of
The self-assessment is lofty, as many them.
corporate slogans often are, but MidJourney’s Ilya Sutskever, cofounder and chief
catchphrase contains quite a bit of truth. scientist of OpenAI, describes the DALL-E
Experimenting with this service as the 2 system as ‘Transcendent beauty as a
persona of a novelist, ie someone who paints service’. OpenAI believes DALL-E 2 will
with words rather than with images, using help inspire artisans of all kinds, including
MidJourney was extraordinarily liberating. painters, graphic designers, architects, book
In my experiments, verbal prompts created illustrators, costume designers, storyboard
palettes of moody, smog-filled and noir- artists as well as many simple amateur artists
inspired streets of 1950s Singapore (see looking to improve their craft. These text-
Figure 2) or mythical heroes in epic to-image solutions can simplify an artist’s or
compositions (see Figure 3) in less than a designer’s life, if only in the art discovery,
minute. sampling and collage phase of the creative
Text-to-image systems work by using ML process.
algorithms instead of human intelligence; A simple prompt of ‘/imagine’ triggers
they try different combinations of words MidJourney, telling it to pull images from the
until they come up with something that web and then, according to any provided text
looks like an image or sounds like human instruction, build accordingly. The system
speech. They also need lots and lots of starts by providing the user with four images,
examples (known as training data) so they any of which can be used as a jumping-off
can learn how language works in general point for more images. An image the user
before being asked questions about specific likes can be created in high resolution.

Figure 2: MidJourney text-to-image landscape prompted by the following text prompts: noir Singapore alley
in 1950, smog, dark, street lights, rule of thirds, cityscape, matte, VHS aesthetic, smoke coming from ground,
amber lights, highly detailed, extra details, clean, trending on ArtStation (ArtStation.Com is an online platform for
creative collaboration and creative showcases)

6   Journal of AI, Robotics & Workplace Automation Vol. 2, 2 1–14 © Henry Stewart Publications 2633-562X (2023)

Pearson.indd 6 30/01/2023 14:56

 The rise of CreAltives: Using AI to enable and speed up the creative process

Figure 3: MidJourney text-to-image portrait prompted with: Epic composition, a boy fights atop a mountain,
highly detailed, torchlight, night, fog, rain, lit windows, night, shape language, bounce light, cool colours,
reflections, rim light, complementary colours, volumetric light, shadow shapes, rule of thirds, concept art,
atmospheric perspective, cinematic, octane-render, photo-real, CryEngine, V-ray, 8k, micro details, in the style
of Peter Mohrbacher, Edmund Dulac, Greg Rutkowski, muted color scheme::0, by simon bisley::0, by greg
rutkowski::0, by lee bermejo::0, by alphonse mucha::0, by enki bilal::0, by frank frazetta::0, by alex ross::0, by
frank miller::0, by todd mcfarlane::0, by dave mckean::0, by ayami kojima::0, by shinkiro::10000, by masamune
shirow::0, by hirohiko araki::0, by takehiko inohue::0, by luis royo::0

To produce Figure 3, we experimented as a whole paragraph of instructions and the

the persona of a visual artist. We input a list results proved impressive. Of course, a
of artists we wanted the work to be inspired close-up inspection reveals eyes that do not
by, along with mood settings, ray tracing look quite right, but this is nit-picking.
information and more. Some prompts, Images like these can trigger a whole host of
like ‘boy fights atop a mountain’, were new ideas for a character or a book or movie
completely ignored, but there is enough in plot. A text-to-image AI system can capture
the image for me to use as an inspiration for the essence of what a writer or an artist is
a character sketch. The prompt contained looking for in a character or in marketing

© Henry Stewart Publications 2633-562X (2023) Vol. 2, 2 1–14 Journal of AI, Robotics & Workplace Automation   7

Pearson.indd 7 30/01/2023 14:56

 Pearson

collateral with only a text and/or an image the mind of the user should pop up. If not,
prompt, all in average time of less than 60 well, there is always a tweak of the prompt
seconds. and then another 60-second wait for several
By applying techniques such as style newly rendered images that might just capture
transfer and other ML-trained methods what you were looking for in your prompt.
to generate visuals that would have been One does have to trust in the system,
very time-consuming to produce even for however. ‘The creative process is a process
a skilled professional, AI can lower the of surrender, not control’, Bruce Lee once
barrier of entry to an extent that may lead said. That is a motto to keep in mind while
to the devaluation of skills and expertise working with MidJourney, which effortlessly
of creatives. These technologies will likely seduces you down a rabbit hole of creativity
improve significantly in the future and could that is highly addictive and endlessly creative.
have a distinct impact on the way creatives It does not decipher dreams, but it sure lets
express and manifest their talents.37 The user your imagination run wild. Some control
can freely create anything their imagination must be relinquished, but as OpenAI’s
desires, in any chosen artistic style they fancy, Sutskever asserts, every now and then
from the standard styles of Art Nouveau, MidJourney generates something that just
Cubism, Expression and Modernism to makes you gasp.
Post-Impressionism. The palette can go from In his article, ‘AI-generated imagery is
Afrofuturism to Ziggurat, with anything and the new clip art as Microsoft adds DALL-E
everything in between. With MidJourney to its Office suit’,38 James Vincent explains
and DALL-E, the world of art and the style that ‘Microsoft is adding AI-generated art
of any known artist, even highly obscure to its suite of Office software with a new
ones, are available at the click of a button. app named Microsoft Designer’. The app
Because MidJourney works on a Discord works like other AI text-to-image tools,
server, everyone’s work is accessible to with user prompts generating a variety of
all who are logged onto that server, so designs with minimal effort.39 ‘Designer can
collaboration can be a part of the process. be used to create everything from greeting
Anyone can whip off a variation of a piece cards and social media posts to illustrations
they see in the render cue. Watching the for PowerPoint presentations and logos for
generated art is like seeing the history of businesses’, says Vincent.40 Designer will
art playing out on an hourly basis. Using have a free standalone app as well as a more
MidJourney reminds one that art is, above feature-rich version available to paying
all else, a collaborative process, owned Microsoft 365 subscribers.41
by anyone and everyone — at least the Depending on how the style transfer
generated art of the digital kind. feature is implemented, ‘it could have
All MidJourney renderings start with the wide-ranging ripple effects in the creative
prompt ‘/imagine’. As a request is typed in, community’, warns Pfeiffer Consulting.42
the system’s promise that ‘There are endless They further contend:
possibilities …’ pops up and that line certainly
rings true. One’s imagination is the only ‘Making style transfer available as a push-
limiter, although the system can go off on button option could have a considerable
confusing and nonsensical tangents every now impact on how creatives format multiple
and then. If it is available as an image on the elements requiring a consistent style, thus
Internet or you can describe it well enough speeding up their workflow — yet in the
in a text prompt, while adding enough artistic process running the risk of reducing the
qualifiers, weights and styles, something kind of creative accident that is often seen
that at least vaguely resembles what was in as a positive force in the creative process.’43

8   Journal of AI, Robotics & Workplace Automation Vol. 2, 2 1–14 © Henry Stewart Publications 2633-562X (2023)

Pearson.indd 8 30/01/2023 14:56

 The rise of CreAltives: Using AI to enable and speed up the creative process

The feature could also facilitate plagiarism, and a RNN continuously updates itself
a concern repeatedly brought up in the based on its continuing input – hence the
research interviews;44 however, a style transfer recurrent part of the name.’
feature would allow for improvised creative
experimentation unavailable without the help Examples of AI music software systems
of AI.45 One designer put it rather succinctly: include Flow Machines by Sony, Google AI’s
‘Creativity does not necessarily mean using NSynth and Jukebox by OpenAI, which
a tool for its intended purpose.’46 ‘Subverting currently includes over 7,000 songs that
tools and letting creativity run rampant in mimic artists such as Frank Sinatra, 2Pac,
ways unintended by the developers is likely Katy Perry and Bruno Mars. When listening
to lead to results we would have a hard time to the songs, one gets the musical intention.
imagining now’, argues Pfeiffer Consulting.47 Other AI applications in music include
assisting sound design and searching through
large databases to find the most appropriate
MUSIC AND AI musical match for a particular application.50
Near the end of the 2010s, one of the
authors of this paper walked through a
Manila (Phillipines) casino and listened to a VIDEO AND AI
casino executive explain how the songs we Anantrasirichai and Bull see AI enhancing
were listening to was part of a soundtrack and helping video postproduction with
customised for the patrons on the floor. colourisation, super-resolution, deblurring,
Using data collected from their patron loyalty denoising, dehazing, turbulence removal and
cards, the casino figured out the average age inpainting, as well as with visual effects.
of the people filling their floor, then adjusted For Matt Cimaglia, an award-winning
the soundtrack accordingly, playing hits from creative director and entrepreneur, it does
whatever generation was most representative not stop there. In his article’ The Future of
on the floor. It was a subtle but clever way to Video Advertising is Artificial Intelligence’,51
use patron data to make the customers more Cimaglia claims, ‘We are witnessing a
amenable to gambling their money away. moment in video marketing history where
Today, applications of AI in music are human editors are becoming obsolete’.
much more sophisticated than building Cimaglia envisions an advertising landscape
soundtracks based on age averaging. ML completely different from our own. In this
algorithms can analyse data to find musical world, advertisers do not shoot a single
patterns, then make suggestions for newly video, they shoot multiple iterations of it.
composed melodies that might inspire an In one, the actor changes shirts. In another,
artist.48 In his article ‘A beginner’s guide the actor is an actress of Asian descent. In
to AI: Neural networks’,49 Tristan Greene another, she is African American. After
explains the process of making music with finishing the shoot, the agency passes the
AI: footage to an algorithm, not a human
editor.52
‘Let’s imagine an AI that generates Instead of taking hours or even days to
original musical compositions based on cut a new ad, the AI algorithm can compile
human input. If you play a note the AI hundreds of videos in a few minutes, each
tries to “hallucinate” what the next note tailored to a specific viewer based on highly
“should” be. If you play another note, the detailed user data.53 Cimaglia contends:
AI can further anticipate what the song
should sound like. Each piece of context ‘As the video analytics flows in, the
provides information for the next step, algorithm can edit the video in real-time,

© Henry Stewart Publications 2633-562X (2023) Vol. 2, 2 1–14 Journal of AI, Robotics & Workplace Automation   9

Pearson.indd 9 30/01/2023 14:56

 Pearson

too—instead of waiting a week to analyze talking head. The system took several images
and act on viewer behavior, the algorithm of a person, ran them through an off-the-
can perform instantaneous A/B tests, shelf ‘face landmark tracker’ to decipher
optimizing the company’s investment in a where the eyes, eyebrows, nose, lips and
day.’54 jawline were, then did the same for another
‘driving’ source video, this time tracking the
Human editors are going extinct, says motion frame by frame.61 Blain explains:
Cimaglia.55
This is personalisation marketing on ‘These networks start out really bad at
a scale never seen before, only available their jobs, but as they perform their jobs
because of AI. Content is tailored to the millions of times, they begin to improve,
subjective individual, not the general, barely and the competition between the two
understood mass. Video marketing will be networks is what drives both to continue
surgically striking highly relevant offers to a getting better. The Discriminator network
market of one, not firing a shotgun blast of isn’t looking for the same things a human
promotions to the uninterested many. fake-spotter might be looking for, but it
Savvy advertising agencies need to doesn’t matter – whatever it’s looking for,
embrace AI today. The same logic backing it keeps getting better at discriminating, so
programmatic banners and search advertising the Generator network has to keep getting
supports ML and chatbots; computers can better to keep fooling it.’62
just do some things faster, cheaper and more
accurately than their human counterpart.56 Meta (née Facebook) is also getting into the
‘In this future of data-driven dynamic text-to-video game. On 29th September,
content, viewers’ information is siphoned to 2022, Meta announced Make-A-Video,63
AI that determines aspects of the video based ‘a new AI system that lets people turn text
on their data’, explains Cimaglia.57 prompts into brief, high-quality video clips’.
The options for customisation do not Facebook has been one of the leaders in AI
just stay with user data. If it is raining where these past few years and its Make-A-Video
the viewer is, it could be raining in the builds upon its past work in generative
video, which is easily done by the agency technology research, which it believes has
plugging in a geolocating weather script.58 the potential to open new opportunities
Cimaglia is correct when he contends, ‘this for creators and artists alike.64 ‘With just a
customization model of video production few words or lines of text, Make-A-Video
is more effective than the current model can bring imagination to life and create
of creating a single video for the masses’.59 one-of-a-kind videos full of vivid colors,
Although he argues there is always a place characters, and landscapes’, says Meta.65 The
for the multimillion-dollar, 30-second Super system generates videos from images or takes
Bowl mini-film, marketers need to get more existing videos and creates similar ones.66
sophisticated when it comes to marketing to Meta is open-sourcing its generative AI
the individual. research to build a supportive community,
It is not just in the editing room where whom they hope will provide useful
AI will make a big difference. AI systems are feedback to improve their product. This is a
now turning single images into talking head common procedure with AI tools.
videos. As Loz Blain explains in his article,
‘Samsung AI brings the Mona Lisa (or any
other picture) to life’,60 Samsung’s AI Center ANIMATION AND AI
in Moscow uses adversarial learning to take AI has been widely used in the gaming
a single image of a person and turn it into a industry. Immersive VR and mixed reality

10   Journal of AI, Robotics & Workplace Automation Vol. 2, 2 1–14 © Henry Stewart Publications 2633-562X (2023)

Pearson.indd 10 30/01/2023 14:56

 The rise of CreAltives: Using AI to enable and speed up the creative process

(MR) experiences require good quality, ‘AI Dungeon is a web-based game that is
high-resolution, animated worlds, which capable of generating a storyline in real-
pose new problems for data compression and time, interacting with player input. The
visual quality assessment.67 AI technologies underlying algorithm requires more than
have been used to make this immersive 10,000 label contributions for training to
content more exciting and realistic. It helps ensure that the model produces smooth
robustly track and localise users and the interaction with the players.’82
objects they see and interact with inside
a virtual environment.68 For example, Modern games often incorporate 3D
Meta’s Oculus Insight uses AI to generate visualisation, AR and VR in an attempt to
real-time maps and position tracking.69 make play more realistic and immersive.83
‘Combining audio and visual sensors can Some games even generate synthetic 3D
further improve navigation of egocentric gaming environments with the help of
observations in complex 3D environments’, deep neural networks trained on real video
say Anantrasirichai and Bull.70 cityscapes.84 DL technologies have also been
Recently, AI has been used to make the used in VR/AR game design85 and emotion
animation process faster, simpler and more detection has even been added to improve
realistic than in the past.71 ML is particularly a game’s immersive experience.86 ‘Recently
good at creating models of motion from AI gaming methods have been extended into
captured real motion sequences.72 Google the area of virtual production, where the
research has created software for pose tools are scaled to produce dynamic virtual
animation based on PoserNet and FaceMesh environments for filmmaking’, contend
that turns a human pose into a cartoon Anantrasirichai and Bull.87
animation in real time.73 ‘Get into character’, When it comes to rendering objects and
asserts the Adobe tagline for its Character scenes, AI has also proven instrumental for
Animator software, a solution that allows a gaming, helping with the synthesis of 3D
user to create a 3D character animation that views from motion capture, ray-tracing
replicates their moves.74 The software, which lighting, character and scene shading,88
synchronises ‘live-performance animation dynamic texture synthesis89 and multiple
with automatic lip sync and face and body depth sensors.90
tracking’, has been embraced by Hollywood
studios as well as many livestreaming content
creators.75 Facebook’s Reality Labs has also FINAL TAKEAWAYS ON CREATIVITY
‘employed ML-AI techniques to animate AND AI
realistic digital humans, called Codec Avatars, ‘The threat some see in the impact of AI and
in real time using GAN-based style transfer machine learning on human creativity is not
technology’.76 new. The history of computer-based visual
Beyond animation, gaming ‘could be creation is shaped by the democratization of
considered as an “all in-one” AI platform, tools, making the production of sophisticated
since it combines rendering, prediction and designs ever easier and more accessible’, says
learning’, say Anantrasirichai and Bull.77 Pfeiffer Consulting.91
It has supported design, decision making Starting with the simple page layout and
and interactivity,78 helped with interactive vector design tools of the 1980s and 1990s, the
narrative builds,79 assisted in generating evolution has been extremely liberating, but its
procedural content80 and deep reinforcement has left many creative professions in its wake.92
learning has been employed for in-gaming Currently, there is a widespread fear that AI
personalisation.81 As Anantrasirichai and Bull will have a leveling effect on creative output,
explain: leading to a homogenised, machine-driven

© Henry Stewart Publications 2633-562X (2023) Vol. 2, 2 1–14 Journal of AI, Robotics & Workplace Automation   11

Pearson.indd 11 30/01/2023 14:56

 Pearson

mediocrity that allows almost anyone to of what creatives need to demonstrate

produce impressive, ad-ready results.93 The ever more strongly: That they, unlike the
question is, will this make the differentiation machine, can fully focus on the WHY
between machine-produced output and the of a project, while the machine can only
work a creative professional provides harder to produce the WHAT.’97
perceive, especially to the untrained eye?94 ‘AI
will shift the designer to be a creative director. The ancient Greeks gave the gods too much
That will shift the focus from creating to credit when discussing creativity and maybe
decision-making — but no AI will supplant our current generation, a few years down the
human decision-making’, believes Evan line, will be giving too much credit to AI.
Abrams, a Motion Designer and Instructor.95 Seeing what is available with the tools now,
The act of creation has always been one however, it boggles the mind to think about
of seeking out inspiration so that we can tell what the future holds. The American author,
a narrative that elicits powerful emotions in Dorothy Parker, once said, ‘Creativity is a
a receptive audience. Novelists do it with wild mind and a disciplined eye’. Humans
words, painters with colour, photographers keeping the wild while handing off the
with lights and lenses, composers manipulate discipline to AI sounds like a deal most
it with music and silence, actors with voices, creatives would give an arm and a leg to
expressions and movement, filmmakers do take. But maybe the American author, Jack
it with all of the above and more. With AI, London, had a point too when he said, ‘You
new tools have been added to the creative’s can’t wait for inspiration, you have to go after
arsenal that will help them illuminate and it with a club’. Perhaps, today, there is an
disseminate their message. easier way to catch that creative spirit — just
We are a long way off from AI producing follow the muse that goes by the name of ‘AI’.
anything consequential as ground-breaking
art, as much as GANksy (AI program trained
on Banksy’s street art)96 might want to try, References
but AI can act as a graphic designer’s mood 1. Anantrasirichai, N. and Bull, D. (July 2021),
board, an artist’s sketch pad, a costume ‘Artificial intelligence in the creative industries: A
review’, Springer, available at https://link.springer.
designer’s lookbook, a book cover illustrator’s com/content/pdf/10.1007/s10462-021-10039-7.pdf
outliner or a film director’s storyboard artist. (accessed 9th October, 2022).
At the very least, it is a wonderful tool 2. Ibid.
3. Ibid.
for any budding artist who wants to learn 4. Ibid.
about the concepts, themes and history of 5. Ibid.
art, photography and/or cinema. There is a 6. Ibid.
7. Ibid.
stream-of-consciousness exuberancy about 8. Pfeiffer Consulting (2008), ‘Creativity and
the tool that provokes the imagination in technology in the age of AI’, ResearchGate,
ways both wonderful and enriching. available at https://www.researchgate.net/
Anantrasirichai and Bull conclude: publication/338840672_Creativity_and_technology_
in_the_age_of_AI (accessed 14th October, 2022).
9. Ibid.
‘[And] these are also the core aspects that 10. Ibid.
will thrive as AI and ML progress in the 11. Ibid.
12. Ibid.
creative space: Creatives are aware of what 13. Caramiaux, B. and Donnarumma, M. (July 2020),
they can achieve. Those core human ‘Artificial Intelligence in Music and Performance:
attributes of creativity have always been A Subjective, Art-Research Inquiry’, available at
https://arxiv.org/pdf/2007.15843.pdf (accessed 14th
their strongest asset: Now we can expect October, 2022).
these qualities of vision, connection and 14. Ibid.
empathy to become the key component 15. Ibid.

12   Journal of AI, Robotics & Workplace Automation Vol. 2, 2 1–14 © Henry Stewart Publications 2633-562X (2023)

Pearson.indd 12 30/01/2023 14:56

 The rise of CreAltives: Using AI to enable and speed up the creative process

16. Pfeiffer Consulting, ref. 8 above. 48. Anantrasirichai and Bull, ref. 1 above.
17. Ibid. 49. Greene, T. (July 2018), ‘A beginner’s guide to
18. Ibid. AI: Neural networks’, TNW, available at https://
19. Ibid. thenextweb.com/news/a-beginners-guide-to-ai-
20. Ibid. neural-networks (accessed 5th October, 2022).
21. Ibid. 50. Anantrasirichai and Bull, ref. 1 above.
22. Ibid. 51. Cimaglia, M. (December 2018), ‘The Future
23. Ibid. of Video Advertising is Artificial Intelligence’,
24. Ibid. Entrepreneur, available at https://www.entrepreneur.
25. Ibid. com/article/323756 (accessed 15th October, 2022).
26. Davies, J., Klinger, J., Mateos-Garcia, J. and 52. Ibid.
Stathoulopoulos, K. (June 2020), ‘The Art in 53. Ibid.
the Artificial: AI and the creative industries’, 54. Ibid.
Nesta, available at https://cdn2.assets-servd.host/ 55. Ibid.
creative-pec/production/assets/publications/PEC- 56. Ibid.
and-Nesta-report-The-art-in-the-artificial.pdf 57. Ibid.
(accessed 14th October, 2022). 58. Ibid.
27. Ibid. 59. Ibid.
28. Ibid. 60. Blain, L. (May 2019), ‘Samsung AI brings the Mona
29. Ibid. Lisa (or any other picture) to life’, Newatlas, available
30. Caramiaux, B., Lotte, F., Geurts, J., Amato, G., at https://newatlas.com/samsumg-ai-mona-lisa-
Behrmann, M., Falchi, F., Bimbot, F., Garcia, A., photo-talking/59828/ (accessed 16th October, 2022).
Gibert, J., Gravier, G,. Holken, H., Lefebvre, S., 61. Ibid.
Liutkus, A., Perkis, A., Redondo, R., Turrin, E., 62. Ibid.
Vieville, T. and Vincent, E. (2019), ‘AI in the media 63. Facebook (September 2022), ‘Introducing Make-A-
and creative industries’, available at https://arxiv.org/ Video: An AI system that generates videos from text’,
pdf/1905.04175.pdf (accessed 14th October, 2022). available at https://ai.facebook.com/blog/generative-
31. Pfeiffer Consulting, ref. 8 above. ai-text-to-video/ (accessed 17th October, 2022).
32. Ibid. 64. Ibid.
33. Ibid. 65. Ibid.
34. Ibid. 66. Ibid.
35. Heaven, W. D. (April 2022), ‘This horse- 67. Anantrasirichai and Bull, ref. 1 above.
riding astronaut is a milestone on AI’s 68. Ibid.
long road towards understanding’, MIT 69. Ibid.
Technology Review, available at https://www. 70. Ibid.
technologyreview.com/2022/04/06/1049061/ 71. Ibid.
dalle-openai-gpt3-ai-agi-multimodal- 72. Ibid.
image-generation/#:~:text=Artificial%20 73. Ibid.
intelligence-,This%20horse%2Driding%20 74. Adobe, ‘Animate in real time. Really’, available
astronaut%20is%20a%20milestone%20on%20 at https://www.adobe.com/products/character-
AI’s%20long,what%20we%20mean%20by%20 animator.html (accessed 17th October, 2022).
intelligence.&text=When%20OpenAI%20 75. Anantrasirichai and Bull, ref. 1 above.
revealed%20its%20picture%2Dmaking%20neural%20 76. Ibid.
network%20DALL%2DE (accessed 15th October, 77. Ibid.
2022). 78. Justesen, N., Bontrager, P., Togelius, J. and Risi,
36. Ibid. S (2020), ‘Deep learning for video game playing’,
37. Pfeiffer Consulting, ref. 8 above. IEEE Trans Games, Vol. 12, No. 1, pp. 1–20
38. Vincent, J. (October 2022), ‘AI-generated imagery 79. Riedl, M. and Bulitko, V. (2012), ‘Interactive
is the new clip art as Microsoft adds DALL-E to its narrative: A novel application of artificial intelligence
Office suite’, The Verge, available at https://www. for computer games’, 16th AAAI Conference on
theverge.com/2022/10/12/23400270/ai-generated- Artificial Intelligence.
art-dall-e-microsoft-designer-app-office-365-suite 80. Héctor, R. (2014), ‘MADE – massive artificial
(accessed 16th October, 2022). drama engine for non-player characters’, FOSDEM
39. Ibid. VZW, TIB AV-Portal, available at https://av.tib.eu/
40. Ibid. media/32569 (accessed 16th October, 2022).
41. Ibid. 81. Wang, P., Rowe, J., Min, W., Mott, B. and Lester,
42. Pfeiffer Consulting, ref. 8 above. J. (2017), ‘Interactive narrative personalization with
43. Ibid. deep reinforcement learning’, International Joint
44. Ibid. Conference on Artificial Intelligence, available at
45. Ibid. https://www.ijcai.org/proceedings/2017/0538.pdf
46. Ibid. (accessed 16th October, 2022).
47. Ibid. 82. Anantrasirichai and Bull, ref. 1 above.

© Henry Stewart Publications 2633-562X (2023) Vol. 2, 2 1–14 Journal of AI, Robotics & Workplace Automation   13

Pearson.indd 13 30/01/2023 14:56

 Pearson

83. Ibid. dynamic texture synthesis’, The IEEE Conference on

84. Ibid. Computer Vision and Pattern Recognition (CVPR).
85. Zhang, G. (2020), ‘Design of virtual reality 90. Guo, K., Lincoln, P., Davidson, P., Busch, J., Yu, X.,
augmented reality mobile platform and game user Whalen, M., Harvey, G., Orts-Escolano, S., Pandey,
behavior monitoring using deep learning’, International R., Dourgarian, J., DuVall, M., Tang, D., Tkach, A.,
Journal of Electrical Engineering & Education, available Kowdle, A., Cooper, E., Dou, M., Fanello, S., Fyffe,
at https://doi.org/10.1177/0020720920931079 G., Rhemann, C., Taylor, J., Debevec, P. and Izadi,
(accessed 16th October, 2022). S. (2019), ‘The relightables: Volumetric performance
86. Quesnel, D., DiPaola, S. and Riecke, B. (2018), capture of humans with realistic relighting’, ACM
‘Deep learning for classification of peak emotions Transactions on Graphics, Vol. 38, No. 36, Art. 217.
within virtual reality systems’, International SERIES 91. Pfeiffer Consulting, ref. 8 above.
on Information Systems and Management in Creative 92. Ibid.
Media, pp. 6–11. 93. Ibid.
87. Anantrasirichai and Bull, ref. 1 above. 94. Ibid.
88. Nalbach, O., Arabadzhiyska, E., Mehta, D., Seidel, 95. Ibid.
H. P. and Ritschel, T. (2017), ‘Deep shading: 96. Azarello, N. (October 2020), ‘AI software GANksy
Convolutional neural networks for screen space was shown street art, and learned to draw like
shading’, Computer Graphics Forum, Vol. 36, No. 4, banksy’, Designboom, available at https://www.
pp. 65–78. designboom.com/art/ai-software-ganksy-banksy-
89. Tesfaldet, M., Brubaker, M. A. and Derpanis K. G. learned-10-28-2020/ (accessed 17th October, 2022).
(2018), ‘Two-stream convolutional networks for 97. Anantrasirichai and Bull, ref. 1 above.

14   Journal of AI, Robotics & Workplace Automation Vol. 2, 2 1–14 © Henry Stewart Publications 2633-562X (2023)

Pearson.indd 14 stats
View publication 30/01/2023 14:56