The Prompt Artists

MINSUK CHANG, Google, Inc., United States

STEFANIA DRUGA, Google, Inc., United States
ALEX FIANNACA, Google, Inc., United States
PEDRO VERGANI, Google, Inc., United Kingdom
CHINMAY KULKARNI, Google, Inc., United States
CARRIE CAI, Google, Inc., United States
arXiv:2303.12253v1 [cs.HC] 22 Mar 2023

MICHAEL TERRY, Google, Inc., United States

Fig. 1. Prompt artists develop descriptive text-based prompts that are rendered by text-to-image models. Highly skilled
prompt artists will develop 1) distinct visual concepts and styles (S1), 2) prompts that can also serve as titles of the art piece
(“prompts as art”, A1), and 3) “prompt templates” (A2), which encapsulate specific visual concepts to be customized by others.
Artists strive to discover unique natural language that produces unique visual outputs (G1), and/or model “glitches” (G2)
that can be elevated to artistic styles in their own right. Finally, some prompt artists validate the novelty of their work by
conducting an image search for similar images (C1).

This paper examines the art practices, artwork, and motivations of prolific users of the latest generation of text-to-image
models. Through interviews, observations, and a user survey, we present a sampling of the artistic styles and describe the
developed community of practice around generative AI. We find that: 1) the text prompt and the resulting image can be
considered collectively as an art piece (prompts as art), and 2) prompt templates (prompts with “slots” for others to fill in with
their own words) are developed to create generative art styles. We discover that the value placed by this community on unique
outputs leads to artists seeking specialized vocabulary to produce distinctive art pieces (e.g., by reading architectural blogs to
find phrases to describe images). We also find that some artists use “glitches” in the model that can be turned into artistic
styles of their own right. From these findings, we outline specific implications for design regarding future prompting and
image editing options.
Conference acronym ’XX, June 03–05, 2018, Woodstock, NY
© 2018 Association for Computing Machinery.
This is the author’s version of the work. It is posted here for your personal use. Not for redistribution. The definitive Version of Record
was published in Proceedings of Make sure to enter the correct conference title from your rights confirmation emai (Conference acronym ’XX),
https://doi.org/XXXXXXX.XXXXXXX.

1
Conference acronym ’XX, June 03–05, 2018, Woodstock, NY Chang et al.

CCS Concepts: • Human-centered computing → Empirical studies in HCI.

Additional Key Words and Phrases: AI art, Artists using AI, Text-to-Image models
ACM Reference Format:
Minsuk Chang, Stefania Druga, Alex Fiannaca, Pedro Vergani, Chinmay Kulkarni, Carrie Cai, and Michael Terry. 2018. The
Prompt Artists. In Proceedings of Make sure to enter the correct conference title from your rights confirmation emai (Conference
acronym ’XX). ACM, New York, NY, USA, 19 pages. https://doi.org/XXXXXXX.XXXXXXX

1 INTRODUCTION
Advances in text-to-image(TTI) models have led to significant improvements in the quality of computer-generated,
synthetic images [19, 30]. A new generation of text-to-image models enable the creation of high-fidelity images
via descriptive text prompts by leveraging advances in large language models [3, 47, 54, 55, 60]. With broadening
access to these models, communities of practice have emerged, enabling people to share designs, prompts, and
example images. For instance, there are now tools to help people write prompts 1 , and even marketplaces for
successful prompts 2 .
Prior work has examined the phenomenon of computer-generated art in a variety of contexts [6, 12, 40]. For
example, as an art historian analyzing the AI-assisted art movement, Mazzone et al. [40] describe how the artist’s
role has adapted to include pre-curation, tweaking, and post-curation. More recently, Hertzmann argues that
text-to-image (TTI) models like DALL·E do not themselves create art, but that the artists and technologists who
apply them as tools are the ones creating art [29]. With the emergence of this new class of models, which are
capable of producing extremely high quality images from textual descriptions (e.g., [2, 4, 47]), we are motivated
to understand how this new technology is being adopted and used by creators.
In this research, we provide a snapshot of a vibrant community of art practice that has arisen around text-
to-image models, sharing insights into the ingenuity and creativity of the users of these models.3 . Within a
US-based technology company that has produced its own TTI models, we sent a survey to the TTI models users
to gain a basic understanding of how and why they are used. We also interviewed and observed 11 prominent
users of these models who are using the models as an art medium, recruiting from survey respondents and by
directly asking prominent users. They have each generated thousands of images with both the internal and
other publicly available models, and are actively sharing their creations in multiple communities. In studying
the artistically-driven members of this local community, we sought to understand their practices, their artifacts,
and their motivations for engaging extensively with the models. For the purposes of this research, we scope our
inquiry to studying interfaces that only accept text as input (recognizing that a wide variety of model capabilities
and interfaces are available, including those that enable more fine-grained editing of images). We restrict our
scope to text-only interfaces because these were the first interfaces available for models such as DALL-E, and
these text-based interfaces have seen considerable use by the public and our internal users.
Our study reveals that users of these models have developed a range of artistic styles, including origami
figurines, fashion (e.g., dresses) made out of materials like bricks, and reality “mash-ups” that create hybrids of
animals or of fruits and vegetables (see Figures 2b, 3a, 3b, 4b). However, we also found that the artistic outputs of
this community of users are not limited to the images themselves. For example, the prompt itself is an important
output, and a piece of the art: a parsimonious, descriptive prompt accompanying the image is seen as a virtuous
goal beyond just the image, as it simultaneously acts as a “title” for, and description of, the art piece (see Figure 1,
A1). Similarly, a prompt template—a text prompt with one or more empty “slots” for others to fill in—is considered
an art piece all on its own (see Figure 1, A2). Among other characteristics, a well-designed prompt template has
1 https://promptomania.com/prompt-builder/
2 https://promptbase.com/shop
3 We thank our anonymous reviewers for the specific phrase recognizing our contribution.

2
The Prompt Artists Conference acronym ’XX, June 03–05, 2018, Woodstock, NY

the property of encapsulating an artistic vision that can nonetheless be customized by future users of that prompt
template.
Our results also reveal the lengths some users go to when searching for unique, distinctive outputs. In particular,
some creators turn to thesauri or online, domain-specific blogs (e.g., architectural blogs) in search of vocabulary
that elevates the model output beyond the ordinary. This focus on vocabulary suggests that capable TTI model
artists may also benefit from being highly skilled with natural language. Another creator explicitly seeks unique
model outputs, but through identification of “glitches” that can be elevated to styles all on their own. For example,
this latter artist found the model did not render reflections in mirrors perfectly, and explored this concept through
a number of pieces (see Figure 1, G2).
Finally, we find that the artists interviewed place a premium on originality, with some turning to image search
to validate that their outputs are, in fact, unique.
In sum, this paper presents results from a survey and interview study of heavy users of TTI models, making
the following contributions:
• Usage summary: From survey data from 161 respondents, we find that when they use a model, 20% of
respondents report using a TTI model for one or more hours at a time, indicating fairly sustained use of
these models by a sizable portion of the community surveyed.
• Sample styles: We provide a sampling of artistic styles developed by study participants to contextualize
the types of outputs being produced by new text-to-image models.
• Prompt as art: We find that the prompt itself is often considered a part of the artistic output (in addition
to the actual image), with artists pursuing a goal of creating parsimonious, descriptive input prompts.
• Prompt templates as art: We discover that artists also produce prompt templates to encapsulate a unique
visual concept that others can customize.
• Natural language mastery for visual language artistry: We describe how TTI artists seek unique
natural language in an attempt to elevate their pieces beyond the norm.
• Glitches as art: We show how some artists look for “glitches” that can be reliably transformed into new
styles.
• Validating originality: We describe artists’ concerns in validating their outputs as original, and how they
currently validate through image search.
Together, these findings suggest new directions for interactive interfaces and aids for prompt-centric uses of
TTI models: 1) Methods and tools to help users locate novel language and capabilities of the model, 2) aiding users
in validating the originality of their outputs, and 3) reifying the notion of a prompt template into a standalone
computational artifact that supports richer interaction by users of the template. Importantly, while our results
derive from a study of internal TTI models, the implications for design are generally applicable to use of any TTI
model (e.g., the notion of a prompt template is useful for any TTI model, as it captures a particular artistic vision
in a portable, yet customizable. form).
In the rest of the paper, we review related work, describe our study method, present results from the survey
and interview study, and conclude with a discussion that draws implications for design from the study data.

2 RELATED WORK
Advances in deep learning have led to the development of generative machine learning models capable of
producing both images that are highly realistic and images that are highly creative in both existing and novel
artistic styles [11]. For example, Generative Adversarial Networks (GANs) [23] learn to generate images and
simultaneously distinguish real and fake images to generate highly realistic images. Many variations of the GAN
architecture have been investigated (e.g., [8, 37, 62]). A particularly relevant variation of this architecture is the
Creative Adversarial Network (CAN) from Elgammal et al. [20], designed to generate images with novel artistic

3
Conference acronym ’XX, June 03–05, 2018, Woodstock, NY Chang et al.

styles. This artwork was subsequently featured in multiple exhibitions [1] where human observers could not
distinguish the CAN-generated art from human-authored artwork. Outside of GANs, Gatys et al. [22] introduced
a method to apply learned artistic styles to random images (a technique now known as Neural Style Transfer
[35]). Additionally, work originally designed to make convolutional neural networks more explainable, now
referred to as Deep Dreams, became popular for generating art [42] due to its ability to generate psychedelic
versions of images [44]. While these techniques allowed for generating images with creative and novel artistic
styles [20], none provided significant affordances to end-users for controlling what was generated outside of the
training data scope.
Mansimov et al. [39] addressed this issue by showing that a generative model could produce novel images
from natural text when conditioned on image captions. As text-to-image models rely on language modeling
techniques, recent advances in the scaling of large language models [18, 51] have enabled the development of
correspondingly large text-to-image models with impressive results. The most recent of these models include:
DALL·E [48, 53] and DALL·E 2 [47, 52], Stable Diffusion [5, 54], Midjourney[3], Parti [4, 60], and Imagen [2, 55].
Fueled by the latest advances in text-to-image models, current image generation applications are becoming
mainstream. With this broader adoption comes the question of how these new models’ capabilities impact art
practices, which we examine in this paper.

2.1 AI in Creativity Support Tools & Human-AI Co-Creation

AI tools have played a prominent role in creativity support tool (CST) research [21, 33]. AI-based CST systems
have been produced to support artistic generation in domains such as fashion and product design [34, 50, 56],
music creation [32, 38, 41], drawing [16, 17, 36, 46], visual design and story-boarding [57, 61], and storytelling
[31, 49]. Most of these tools support the artistic implementation process as either production aids (i.e., tools that
perform most of the work of generating the art, e.g., generative text-to-image models) or as execution aids (e.g.,
AI-powered brush tools in drawing applications) [15].
Hwang et al. [33] further characterize how these tools apply AI models in the creative process as falling
into four general categories: editors (facilitate execution of processes), transformers (aid in changing existing
content), blenders (combine 2 or more content sources), and generators (produce novel content). In this framing,
the large-scale TTI models that this work is focused on fall into the generators category.
Additionally, research in the related field of Human-AI Co-Creation (a sub-field of Mixed-Initiative Co-Creativity
[59]) is highly relevant. In the Library of Mixed-Initiative Creative Interfaces [58], Spoto et al. proposed a framework
to understand mixed-initiative co-creation as a process involving seven potential actions: ideate, constrain, produce,
suggest, select, assess, and adapt. Muller et al. [45] extended this framework to generative AI applications, while
Grabe et al. further simplified this extension and characterized four primary interaction patterns concerning
GAN applications: curating, exploring, evolving, and conditioning [24]. In our work, we observe similar themes,
especially the notion of users of TTI models feeling like they are using the models to explore, and acting as
curators of their outputs.
Work in this field has also identified core challenges in creating human-AI co-creation systems. For example,
Chung et al. [14] identified the limited ability to control the output of generative AI models and proposed
using gestural input to constrain/guide the model output. Likewise, Buschek et al. [10] identified a set of nine
challenges system designers could encounter when developing human-AI co-creative systems. In the context of
TTI models, they identified challenges of invisible AI boundaries (“A (generative) AI component imposes unknown
restrictions on creativity and exploration”) and conflicts of territory (“AI overwrites what the user has manually
created/edited”) as particularly salient. Participants in our study encountered similar challenges of identifying
what the models are capable (and not capable) of, and in building upon prior inputs to the system.

4
The Prompt Artists Conference acronym ’XX, June 03–05, 2018, Woodstock, NY

Building on this prior CST research, we examine the emergent practices and goals that have evolved in concert
with the latest generation of TTI models.

2.2 Generative AI as an Artistic Medium

Given the rapid advancement of AI models for generating images with ever more creative and novel styles,
art historians and technologists have been actively discussing how to conceptualize AI-assisted art in relation
to other artforms. Experts in these communities have disagreed as to whether generative models should be
considered artists in-and-of themselves [40] (as with the famous sale of Portrait of Edmond Belamy auctioned by
Christie’s in 2018 [13]) or whether they should be considered as merely a tool employed by artists. Hertzmann
[28, 29] argues that generative AI models are similar to the camera as it relates to the art of photography: a
tool the enables the art. Hertzmann further theorizes that “art is an interaction between social agents,” and
generative AI models are therefore best considered an agent in this interaction. Agüera y Arcas [6] provides a
similar argument with an in-depth discussion of the similarity between the emerging field of AI-generated art and
photography, particularly surrounding the historical reaction of painters to the introduction of the camera. Grba
[26, 27] provides a framework to critically evaluate art created with a generative AI model. In this framework, he
echoes arguments above, and critiques what he refers to as “the ever-receding artist”: the repeated occurrence of
technologists referring to models as artists, thereby minimizing the contribution of the human who employed
the model to create art. Finally, Browne [9] explored what it means to be an “AI artist” and proposed the framing
of an AI artist as a bricoleur (building upon [25]), saying, “Bricolage is common to generative art, where ideas are
developed through playful experimentation with existing tools and techniques.”
In our paper, we present an analysis of interviews with artists who employ a large TTI model for the generation
of their art, highlighting the work of three of these artists and provide context around their motivations and goals.
In our results, we find thematic alignment with perspectives advanced by Hertmann, Agüera y Arcas, and Grba
above: While the models can produce surprisingly high quality output, dedicated users of these models employ
the models as tools to explore specific themes and concepts. Accordingly, they have intentionally developed
processes (e.g., locating domain-specific terminology) to improve their ability to achieve their individual goals.

2.3 Diffusion and Auto-Regressive Models

Diffusion models (e.g., Imagen [2] or DALL-E [47]) are trained by gradually adding noise to an image, until all of
the image is completely noise. The model then learns to reverse the noising process to generate the original image.
In this way, a diffusion model learns to synthesize an image from noisy images, and is capable of generating
images from arbitrary “noise.” For text-to-image models, the models are also trained (conditioned) on text inputs [?
], allowing the model to produce an image from a noisy image input and text input, where the resulting image
bears a resemblance to the input text.
Autoregressive models, such as Parti [4], treat text-to-image generation as a sequence-to-sequence problem,
akin to machine translation or other language modeling tasks. In the case of TTI models, the “translation” is
from text to image (i.e., text tokens to image tokens).
In our study, participants used both types of models.

3 STUDY DESIGN
To understand current practices, motivations, and goals when using modern text-to-image models, we sent
a survey to internal users of two internal TTI models to collect basic information about their use of these
models (e.g., time spent using the models, motivations, desired capabilities, and prompting strategies). We also
interviewed and observed 11 power users of TTI models (8 identifying as Male and 3 identifying as Female) in a
50-minute study to uncover their motivations and practices. The latter participants were prolific users of one or

5
Conference acronym ’XX, June 03–05, 2018, Woodstock, NY Chang et al.

more TTI models. Models used by the participants in the study are anonymized for review, but are of the same
basic capability as state of the art text to image generation models such as Imagen [2], Parti [4], and DALL-E 2
[47] which are described in more detail in our related work.

3.1 Participants
For the survey, participants were recruited from an internal chat channel dedicated to TTI models (where the
internal chat channel has thousands of members) and internal TTI model mailing lists (with hundreds of members).
From the survey respondents, we identified interview candidates who had reported having created artwork over
more than ten sessions and having spent more than five hours in the previous week using a TTI model. To create
a pool of participants, we recruited eight of these latter respondents, and further recruited three prominent
artists in the internal artist community to participate. These three artists are quite visible in the internal artist
community, and have shared their unique artwork collections within that community. Participants were also
actively engaged in external communities, sharing knowledge, expertise, and artwork. Participants were given a
60 USD gift card for participation.

3.2 Interview structure

The study consisted of four parts intended to understand participants’ practices. Each participant was first asked
to create an image of their choice to allow the researcher to observe their natural practices. In the second part of
the study, participants were asked to reflect on 1) an artwork they were proud of, 2) a piece they found most
successful, and 3) the piece that was least successful. In the third part of the interview, participants were asked to
reflect on someone else’s work by examining only the prompt, and specifically asked to either improve or change
the prompt in their style. In the last part of the interview, participants were asked to discuss envisioned uses for
these text-to-image models.
Interviews were conducted remotely. We recorded the shared screen and automatically generated transcripts
for the interviews. We constrained our focus to interfaces that only use text prompts as input to the models. In
addition, some participants voluntarily shared their collection of generated artwork after the interview.

3.3 Qualitative Data Analysis

For the qualitative analyses, the authors analyzed the video transcriptions and also noted comments on participants
non-verbal interactions. The final corpus of automatically generated transcripts was 164 pages (60614 words).
The first two authors each reviewed the transcripts data independently, looking for ways of explaining the artistic
practices [43]. In this process, the authors separately analyzed each transcript to extracted salient themes, and
independently generated hypotheses and points of discussions [7]. Using these data, all authors participated in
two rounds of interpretation sessions to arrive at the primary themes reported in this paper, and resolve any
discrepancies and disagreements. During the interpretation sessions, authors also analyzed the prompts and the
images created by the study participants to identify unique artistic styles and practices. These sessions were
inspired by existing analysis practices from qualitative media analysis [? ].

4 SURVEY RESULTS
We received 161 responses to the survey. Of these responses, 160 answered the question, “At present, why are
you using the [TTI] models?” Of the responses to this question, 79 (49%) indicated they use the models to create
art, 33 (21%) reported using it as part of their creative work pipeline, and 126 (79%) indicated their use was
curiosity-driven (not work-related).
In the survey, we also asked participants to estimate the length of time they work with a TTI model when
they use one (“When you interact with a model, how much time do you typically spend interacting with it?”).

6
The Prompt Artists Conference acronym ’XX, June 03–05, 2018, Woodstock, NY

We received 157 responses to this question, with 20% of respondents indicating that they use a model for one or
more hours at a time when they use it (11% reporting using it for 1-2 hours at a time, 9% using it for 2 or more
hours at a time), 53% indicating they use a model for 10 minutes to an hour, and 27% reporting use for less than
10 minutes.
When asked about observed strengths and weaknesses of the various models they’ve interacted with, survey
responses indicated a number of desired capabilities, such as the ability to render text in images, the ability to have
more control over spatial arrangements, and the ability for models to handle complex prompts. We also asked
participants for desired capabilities when interacting with the model. Seventy six percent of the respondents
requested a “how to build a prompt” guide, and 75% desired the ability to fork and remix images, especially for
spatial refinement. Seventy six percent of the responses also indicated they would like features like bookmarking,
and the ability to directly share outputs to internal chat groups or social media. As we will see in the artist
spotlights below, there is a clear social component to working with these TTI models for our study participants.
Finally, 63% of the responses desired greater control over the model, such as the ability to assign specific values
to each of the prompt words.
When asked to provide prompting techniques they have learned, common themes for the strategies included
1) producing specific art styles and eras, such as “impressionist style”, 2) use of keywords that describe camera
lenses and aperture (e.g., “DSLR photo”,“3D render”,“24 mm, f8, ISO1000”), and 3) domain-specific terms (e.g.,
“Line Art”, “black and white”).

5 PROMPT ARTISTS: STYLES, MOTIVATIONS, PRACTICES

In this section, we provide a sampling of the vibrant internal TTI artistic community by spotlighting the work of
three highly active creators; Shai Noy, Irina Blok, Dan Smith. 4 . For these three artists, we describe and present
examples of the styles they have developed and summarize their artistic motivations and goals. We then provide
a summary of motivations, styles, and practices observed across the 11 interview participants. In the section that
follows, we describe salient high-level, emergent themes arising from the interviews and observations. We want
to credit the artists whose artwork are featured, and they had expressed the desire to be associated with their
artwork. We use their full name in places where the artwork appears.

5.1 Shai Noy: The Explorer (A1)

Shai Noy is a software engineer with no training in the visual arts or design. However, they have produced
thousands of images with TTI models. The styles developed by this artist include “super macro photography”
images (i.e., extremely close-up views of objects, Figure 2a), and fashion (dresses, suits) made out of unusual
materials, such as wood, grass, brick, or ice (Figure 2b).
Elements of both discovery and community were emphasized as rewarding for this artist, such as being the
first to explore particular concepts and the ability to share discoveries: “Everything is more fun when you can
share it” and, “Art doesn’t live in a vacuum, nobody starts from scratch, everything is based on something else. I
am proud of being able to recognize the potential” (A1).

5.2 Irina Blok: The Art Director (A2)

Irina Blok is a designer who has done visual work for many years using stock images and applications like
Photoshop. This artist’s styles include origami dancers (Figure 3a) and reality “mash-ups,” such as sliced produce
that contains different textures internally (e.g., a sliced head of cabbage that reveals a cross-section of an orange,
Figure 3b).
4 In
the text, we use the terms “artists,” “creators,” and “study participants” interchangeably. We denote the three spotlighted artists as A1, A2,
and A3, and other participants by a participant number (e.g., P1, P2, ..., P8).

7
Conference acronym ’XX, June 03–05, 2018, Woodstock, NY Chang et al.

(a) Super macro photography

(b) Dresses in various unusual materials

Fig. 2. Selections from Shai Noy (A1 - The Explorer)

Driving these styles is a passion for developing “impossible objects, something we haven’t seen before, mashups”
(A2). They also seek to create images that differ significantly from the images it was trained on: “The further
away the generated images are from what it was trained on, the more the satisfaction” and, “they’re [the resulting
images] also defying the rules of its training set, like defying [...] gravity. [..] A creative prompt breaks that” (A2).
Importantly, this artist’s output also includes prompt templates that describe a particular, parameterized image
or visual concept, such as this prompt for generating a house: Audacious and whimsical fantasy house shaped like
<object> with windows and doors, <location>. These prompt templates are carefully crafted to reliably produce
pleasing results for others. We describe this concept more fully in a later section.
Notably, when working with the text-to-image models, A2 considers the model to be akin to an artist itself,
with themselves the art director. In this relationship, A2 acknowledges a certain lack of control: “[I] don’t have
full control, and there’s beauty in this” (A2). At the same time, they also consider the model to be a tool: “[The
model] is a brush [...] you just learn to speak its language” (A2). With these dual views (model as an artist, model
as a tool), they note that “the hardest part is [the] conceptual aspect, being skillful with [the] prompt” and that
“it’s a thought exercise, it’s not a visual exercise” (A2). It’s really about how to make people think like an artist”
(A2). In this latter sentiment, this participant was specifically speaking to the need to think like an artist in
formulating a prompt, as opposed to formulating a prompt as if one was talking to a machine.

5.3 Dan Smith: The Social Commentator (A3)

Dan Smith comes from a background in visual media, and is partially driven by the desire to deliver a message
about the climate crisis, in order to facilitate change and awareness: “[I’m] not just having fun ... but [actually]
making something that has some power” (A3). In working towards this goal, they want to make “something that
you look at, and it makes you feel something” while also making “something that people would want to look at”

8
The Prompt Artists Conference acronym ’XX, June 03–05, 2018, Woodstock, NY

(a) Origami dancers

(b) Reality mash-ups

Fig. 3. Selections from Irina Blok, (A2 - The Art Director)

(A3). A3’s image styles align with these overall goals: They have a number of pieces that put nature in “situations
where you wouldn’t see it” (Figure 4a) and have created numerous “hybrid animals” (Figure 4b). This particular
style—hybrid animals—is also inline with a motivation to create images that “would be hard to [...] visualize or
create, if you were [...] a really skilled Photoshop artist” (see Figure 4).
A3 also heavily considers the quality of the image when assessing its outputs: it must have near expert-level
composition, photorealism, and/or artistry. When describing the artwork they created and how they created
them, they noted, “I think the ones that I pick out are [...] standouts for various reasons, and just [...] like what
I said, [...] composition photorealism, artistic quality” (A3). Consistent with their emphasis on overall image
quality, they have found ways to address undesirable outputs. For example, in generating images that include
animals, they found they needed to adopt a specific strategy to create aesthetically pleasing images: “I would
do ‘Tall Grass’ a lot because early on I discovered that limbs and fingers and paws can get a little wonky” (A3).
The “tall grass” addition was their creative strategy to hide feet or paws and suggests an understanding of model
limitations, but also a sense of how to cope with these limitations.

5.4 Summarizing Motivations, Styles, and Practices

One of the primary motivations for interacting with the models was that participants found them fun—the models’
output quality enabled people to feel creative, and they were generally interested in interacting with this new
class of model. Some people also noted that the model enabled them to engage in their domain interests in a new
way. For example, one participant said that the models allowed them to explore their interest in Swiss trains with
the model, while another found it compelling to try to create new forms of currency (e.g., new types of coins).
One participant used it to create the equivalent of clip art for presentations: “I use [TTI] model when I need some
[...] clipart to use in my presentation, and [TTI] model would be amazing for that...” (P2).

9
Conference acronym ’XX, June 03–05, 2018, Woodstock, NY Chang et al.

(a) Climate change

(b) Hybrid animals

Fig. 4. Selections from Dan Smith (A3 - The Social Commentator)

In comparing the artists spotlighted above to the other participants, one notable difference in practices is that
the spotlighted artists placed a particular emphasis on exploring specific themes in depth (e.g., origami figurines).
The other participants did not pursue concepts with the same rigor or depth.

6 THEMES: ART BEYOND IMAGES, DISCOVERING UNIQUE POINTS, AND VALIDATING

ORIGINALITY
Across our interviews and observations, a number of themes emerged. First, the notion of what constituted the
final artistic output was not always just an image: Some creators consider the prompt itself as part of the art.
Similarly, a prompt template can be considered an artistic output. Second, there was a clear desire for creators to
discover new styles possible with the models. This focus on originality extended to one creator going so far as
to conduct an image search on successful outputs to validate their originality. We unpack these themes further
in this section: 1) Prompt as art, 2) Prompt template as art, 3) Discovering new capabilities of the model, and 4)
Validating originality.

6.1 Prompt as Art

For some participants, the prompt itself was part of the overall art, and thus worthy of attention. For these
participants, it was important to “[create] aesthetically pleasing images” and “[develop] art concepts” that were
inherently tied to prompts. We detail these motivations and behaviors below.
While generating aesthetically pleasing, “glitch-free” images was a common goal of the creators, other goals
were also present in their practices. For some participants, the prompt itself was part of the overall art, and thus
worthy of attention 1) on its own, and 2) as it relates to the image: “It’s part of the aesthetic” (P3), where the prompt

10
The Prompt Artists Conference acronym ’XX, June 03–05, 2018, Woodstock, NY

(a) “a photo of an owl turning into a croissant. (b) “a photo of a bananasaurus rex. a
f2.2” bananasaurus rex has the legs and arms of a
banana and the body and head of a
tyrannosaurus rex. sigma 85mm f2.2. studio
lighting.”

Fig. 5. Artist Ian Fischer (P3) considers (a) as a successful art work given the brevity of the prompt, while (b) a failed attempt
because the prompt is too long to get the image they want

is “like a title of the piece, but you don’t get to choose it independently” (P3). Hinted at in this last comment is
the notion that while a prompt could also serve as a title of the art piece, there is a clear dependency on, and
perfunctory role for, that prompt as well: The prompt serves as the source material for the model generating the
resultant image. Given this dependency, finding a prompt that produces the desired result and that can serve as a
title for the piece can be challenging, but rewarding when it happens.
This same artist (P3) further described the prompt’s role as “communicating the image, and the idea of the
image, and how I got it all at the same time” (P3). This quote sheds light on how art with TTI models can, in
some sense, be considered multimodal (text and image) for both the artist and viewer: The prompt and the final
output combine into a single, mutually-reinforcing, art piece. Seen in this light, one can consider the prompt
as art itself—a well-crafted prompt that creates a compelling image but also accompanies that image, saying
something about the image. See Figure 5a for an example of “prompt as art,” as well a prompt that wasn’t able to
achieve this same level of aesthetic (Figure 5b).
Diving deeper into this theme, we observed that the artists believe the “concept” is a critical component of an
artwork. In an internal blog article, A2 describes, “There’s a common misconception art is largely about drawing
and painting skills. Art is not only about how something looks, it’s about what it says, it tells a story, and has
a concept. Art can surprise, provoke, teach, delight and inspire. Art is not just about drawing, art is a way of
thinking.” A1 also suggested “the unit of shareable artwork is not necessarily a specific image but maybe it’s the
whole exploration of the concept of those images.”
In pursuing “prompt as art,” we observed participants impose different constraints when creating their prompts,
with some wanting it as descriptive as possible, while others attempting to make it as simple as possible. One
participant even discovered delight in their accidental discovery that their “random string” produced beautiful
output, and named that prompt as their proudest achievement. In this latter case, the pride comes from the joint
pairing of the random string and the beautiful output—without the context of the random string, the image has
less value, as the random string reveals an unexpected feature of the model.

11
Conference acronym ’XX, June 03–05, 2018, Woodstock, NY Chang et al.

Fig. 6. Example of the prompt “Audacious and whimsical fantasy house shaped like <object> with windows and doors,
<location>”. All images above use “strawberry” as the shape, with the location varied (“countryside”, “Paris”, “Tokyo”).

6.2 Prompt Templates As Art

In a spirit similar to “prompt as art,” five artists sought to produce prompt templates. A prompt template is an
image description with “slots” for someone else to fill in. For example, we previously noted this prompt template
created by A2: Audacious and whimsical fantasy house shaped like <object> with windows and doors, <location>
(see Figure 6 for example outputs of this prompt template).
These prompt templates leverage the capabilities of the models as well as the lightweight, accessible features
of prompts. More specifically, when an artist identifies a compelling composition, they can create a text prompt
that allows others to create a similar composition, but with their own unique customization to it. The ease with
which the templates can be shared also introduces social motivations for producing and distributing templates
(e.g., to participate and contribute to a larger community of practice). We elaborate on these points below.
Prompt templates have a number of key features:
• They are tightly coupled to and represent a particular artistic concept, vision, or composition (such as
a “whimsical fantasy house”). These features make prompt templates conceptually richer than words or
phrases used to specify stylistic characteristics of the image (e.g., “35mm” or “watercolors”, which may be
used to produce a particular effect, but don’t specify a larger composition).
• Others can make use of prompt templates by filling in the blanks. The richness of the models means that
the templates guide the overall generated image, but users’ unique input can yield a diverse variety of
outputs.
• There is the intent for the prompt template to provide consistently high quality and delightful output when
used by others.
Unpacking these concepts in the context of the example prompt above, the skeleton of this prompt embodies a
particular (visual) concept: A fantasy home in a given location. Someone making use of this prompt can customize
it through two key variables: A shape for the house and a location for the house. While these are seemingly
simple variables to customize, the template gives the user great flexibility in terms of the final outputs produced
by the model: Any number of shapes can be provided in any number of locations (in fact, the user is free to
substitute any text in the slots they wish).
Simultaneously, the prompt cues the model to the types of output to produce, as well as details to guide the
generation. The phrase “audacious and whimsical fantasy house” defines desired attributes of the house, while
the specification of “with windows and doors” provides additional details that should be included in the generated
house. These details in the prompt help increase the reliability of the model output and reduce the likelihood that
the downstream user of the prompt template needs to experiment further with the overall prompt structure and
content to obtain a good output.

12
The Prompt Artists Conference acronym ’XX, June 03–05, 2018, Woodstock, NY

A noteworthy characteristic of these templates and the text-to-image models is the rich interplay that results
between the original template and the phrases the user substitutes in the template. For example, a particular
choice of location can profoundly influence the resulting house generated by the model—the model does not
simply generate the same type of house and situate it in a different location. Instead, the choice of location can
directly interact with the other parts of the prompt. For example, creating a strawberry house in the “countryside”,
“Paris”, or “Tokyo” yields qualitatively different outputs for the house, with the house style meshing more naturally
in the chosen location (e.g., having styles of windows more typical for a European building for the house in Paris,
and doors more typical of Japan for Tokyo—see Figure 6). These interactions between the template and the users’
choices enable a diverse variety of outputs that allow a user to explore a wide range of ideas.
To produce prompt templates, one participant described a process whereby they would 1) input a prompt,
2) identify high-quality outcomes, and 3) rewrite the prompt to try to produce that same outcome again. This
process could involve several iterations until they get a reliable prompt. Once a prompt is working, participants
would sometimes remove content to make it more succinct. For example, they would first emphasize a specific
characteristic (like “high resolution”) by applying many descriptors representing a similar effect (such as “high
res”, “DSLR”, “crystal clear”, “photo realistic”), then start removing the repeated qualifiers in the prompt until the
prompt could reliably generate similar outputs to the original, verbose version.
As with the notion of a “prompt as art,” the creation of these prompt templates can also be considered an
artistic outcome in and of itself: The prompt author must first develop a compelling concept, then ensure it has
enough capacity to enable people to produce their own unique creations within the frame of the concept. When
done well, a prompt template has the quality of attracting users (because of the compelling output produced by
the prompt) and continually delighting users with the outputs produced using their unique input.
Elaborating further on this notion of “prompt template as art,” a prompt template represents a particular artistic
vision, with the prompt text capturing the overall design, composition, and aesthetic intentions of the author.
Notably, because the artistic vision is captured in natural language text, a prompt template can be used across
TTI models: Users can customize the template as desired, then tweak the prompt to produce the desired output
using whatever model they have at their disposal.

6.3 Discovering Unique Points in the Model’s Latent Landscape

A common goal of many artists was to discover new capabilities of the model that others had not yet found.
As one participant put it, they tried to “break” the model, pushing it to its limits. As with prompt templates,
these newly discovered capabilities are often shared so others can apply the concept in their own images. For
example, an artist may identify the ability of the model to produce physical sculptures out of unusual materials
or to create origami-like figurines (as A2 did). Once a concept and ability have been identified, others can build
on the core idea and create their own images or permutations. In particular, participants felt proud when they
“made the model do X (new concept)”, especially when they discovered a model capability for the first time in the
community in which they heavily engage, as the discovery could influence the discourse in the community. We
expand on these points below.
new words to steer the model in new directions. For example, one artist mentioned referencing architectural
blogs to learn that domain’s vocabulary so it could be applied to their prompts (A2). Another artist mentioned
turning to a thesaurus to enrich their vocabulary for the prompt. As one concrete example, A1 (the “Explorer”)
likes to employ “unusual words” to steer the model, such as “intricate” instead of “detailed”, explaining: “If you
choose common words, then you get a bit of an uninspiring result quite often. But if you use something a bit
more unusual, then you really narrow down [...] the set [of images], and you’re going to get into things that use
this less common word” (A1).

13
Conference acronym ’XX, June 03–05, 2018, Woodstock, NY Chang et al.

Fig. 7. Artwork by Paul Emmerich(P8). Examples of “breaking the model”: the artist seeks out and elevates “glitches” in
the model’s output to create new styles. The artist takes advantage of the model’s imperfect reflections (on the left) while
making creative use of its tendency to transform the backs of cats into cat faces (see the bottom part of the cat’s back, which
looks like a cat’s face).

What’s noteworthy here is that in the pursuit of novel imagery, creators would carefully research and choose
words to produce specific visual outcomes: What you say and how you say it is critical to producing high-quality
imagery with text-to-image models, requiring TTI users to enrich their natural language vocabulary in order to
develop and skillfully execute a unique visual language.
Driving these practices of discovering new capabilities was a clear desire to push the model away from “average”
outputs and get it into more unique spaces. In this sense, the artists are navigating and charting the vast latent
space of the model and sharing back the most interesting places discovered. When a new, unique output space
was discovered, artists expressed a certain satisfaction in having discovered that space: “ A good rule of thumb is
to be more descriptive than not [...] if you see that something is lacking, then you try to add more descriptors that
will encourage the model, in this direction [...] sometimes you just have to reword a sentence or move something
from one sentence to another [...] it’s mostly like binary search” (A1); “ I think your choice of vocabulary is very
interesting because you want a large tree. But then, instead of saying a large tree you say mature tree [...] Where
do these [...] choices come from? They just come from trial and error” (A2).
In seeking unique spaces, one participant (P8) expressly hunted for interesting imperfections; instead of avoiding
glitches, they sought glitches. For example, this participant found one of their prompts created imperfections in
its output for “a hybrid of a clock and a snail on an infinite mirror. Steampunk. DSLR photo. astrophotography”
(Figure 7). Here, A4 explores the imperfections of the infinite mirror: it’s “technically a failure but still amazing”
(P8). P8 also discovered that the model sometimes has trouble generating the backs of things (like cats) and
developed that behavior into its own art style (Figure 7).

6.4 Validating Originality

One question that often arises on the topic of human-in-the-loop, AI-generated art is one of creativity and
originality—how much can be attributed to the AI versus the person [29]. Our participants also struggled with this
question, with some seeking to ensure their outputs were novel. Participants wanted to validate the originality of
the artistic concept, but could only check the originality of the artifact. For example, one participant described a
practice of using Google’s image search on compelling outputs to ensure originality: After producing a creative
output, they use image search to search for that image (or similar images) to ensure it is, in fact, original. This
participant mentioned they do this in part because they are sensitized to the fact that machine learning models
can sometimes memorize portions of the training data. Given this, they want to ensure their output is not in the
training data.

14
The Prompt Artists Conference acronym ’XX, June 03–05, 2018, Woodstock, NY

7 DISCUSSION
In this research, we have provided a snapshot of the emerging artistic scene enabled by the latest generation of
text-to-image models. Our interviews and observations document the types of imagery artists are developing
with these new models, as well as an enhanced understanding of the types of results creators seek beyond the
images themselves (e.g., prompts and prompt templates as important artifacts in their own right).
In this section, we discuss some implications deriving from artists’ goals of seeking novelty, validating
originality, and producing reusable prompt templates. In reviewing design implications for these models, we note
that there are countless ways the input and editing interfaces could be improved for interacting with these models
(for example, one can find many proposals and working demos online). In this discussion, we restrict ourselves to
interfaces that only take text as input, with no post-processing of the image (such techniques that allow in-filling
of regions by the model). We limit our discussion to this input modality in part because a number of the artists
interviewed were seemingly attracted to the simplicity (and challenge) that this single input modality provides.

7.1 Aiding Novelty

In our interviews, the use of thesauri and domain-specific blogs (e.g., architectural blogs) illustrate the desire of
artists to identify unique terms that help them produce results that rise above the average. Embedding these
search capabilities directly into the tooling could be useful (e.g., quick access to a thesaurus or an embedded
search engine). Pushing this idea further, there may also be opportunities to make use of large language models
(LLMs) and/or the TTI model’s training data to help surface salient terms for a given topic. For example, an
LLM may be able to generate terms-of-art for architecture using a prompt like this: Here are some terms specific
to describing the architectural design of a house: 1). In our own test with an LLM 5 , this prompt yielded terms
like entryway, foyer, entry hall, and elevation. To use the training data itself to identify new spaces, it may be
possible to collect terms associated with a particular topic, such as “house,” then identify other terms that are
more frequently found near that topic in the training data compared to the rest of the training data (e.g., using a
method such as term frequency-inverse document frequency (TF-IDF)).
To help users understand how unique their word choices are, one could also visualize the input prompt with
respect to how common each individual input token is. For templates, one could also show what the common
terms would be for filling in the blanks to help people move beyond those common terms to find more distinct,
unique inputs. One possible outcome in helping people find more novel inputs is that their inputs may lie
outside of the training distribution, leading to unpredictable results. Providing feedback through mechanisms
like visualizations (e.g., showing frequency in the training data) could help users better understand these types of
issues should they arise.

7.2 Validating Originality

As mentioned, there is a desire to validate the originality of the image produced by the model. Streamlining the
process of doing an image search with an online search engine is one obvious way to address this issue. However,
as is the case in aiding novelty (described above), there may be opportunities to take advantage of the training
data itself. More specifically, in addition to external search, one could also search for the closest images in the
training data to the image produced.
While the above mechanisms would help validate the outputs generated, there may also be opportunities to
help validate the originality of the input. For example, one might be able to search the training data to identify
which parts of a prompt exist within the training data.

5 Specific model anonymized for review

15
Conference acronym ’XX, June 03–05, 2018, Woodstock, NY Chang et al.

7.3 Materializing Prompt Templates

Prompt templates provide a way for an artist to derive a new style, then share it with others so they can produce
their own unique images. One could imagine embracing this practice and transforming a prompt template into
its own first-class interface.
For example, one could imagine allowing users to provide multiple inputs for each slot of a prompt template,
then generating the cross-product of all the inputs. One could also make use of the fact that the prompt text
for the templates is tokenized into vectors. Specifically, by supplying two different inputs for the same slot,
embedding vectors for each input could be obtained and then automatically interpolated between the embeddings
to produce a spectrum of outputs. For example, for the shape of a house (in the previous house prompt template),
the user could provide two inputs: strawberry and apple. The system could then produce the embedding vectors
for those inputs, then interpolate between those embeddings to create a series of images that morph from a
strawberry to an apple. However, one thing to keep in mind with these interpolations is they are occurring in the
text space rather than image space—the model will be interpolating not between shapes (per se) but between the
linguistic concepts of strawberries and apples (which may still produce an interesting morphing between these
conceptual entities).

7.4 Prompts and TTI Models as an Art Medium

One the primary ways the three spotlighted artists (A1, A2, and A3) distinguished themselves from others was
their perception of TTI models as an art medium, with a clear focus on exploring the capabilities and limits
of medium itself, rather than only on individual outcomes. In this spirit, they embraced the limitation of not
being able to edit the images directly, accepting the text-only input as a defining feature and characteristic of the
medium. For example, A2 expressed “there’s beauty in it” when describing the prompting interaction with the
models. In contrast, other participants focused more on the outcome, and expressed desire for additional features,
such as direct editing of the generated images. Embracing the TTI models as-is further reinforces the idea that
text prompts are part of the artwork, rather than simply a means to an end.
This observation suggests that the design implications for TTI models can be considered from multiple
perspectives: prompt-only artists, and creative professionals using the models to achieve specific goals. Creative
professionals with a specific design goal may require and request specific features that offer more fine-grained
control, perhaps with tools to help understand model behavior. For example, features related to directly editing
the generated images were among the most frequently requested features in the survey. Given that non-design
experts can also learn to use TTI models to quickly demonstrate and visualize artifacts, we hypothesize that
this may facilitate more active and iterative communication between design professionals and their clients, with
clients more directly engaged in the creative process (as opposed to the more traditional pipeline-like design
workflow). If this proves to be true, feedback and collaboration features could become more important.

7.5 Limitations
Our participant pool was drawn among employees of one large US-based corporation, and does not cover other
possible ways that culture, community, and collaboration might shape use of TTI models (e.g. on social media). Also,
since our analysis was episodic rather than longitudinal, we do not document how artistic prompting strategies
may evolve within individuals. For the interactions we could observe, observing participants’ interactions with
the TTI model does not definitively indicate their conceptions of how the model works or how best to prompt it.
We also acknowledge that different models behave in different ways due to their structure, training data, and the
design of the interface. In that regard, we also acknowledge some of the findings might be model dependent, and
are specific to the models used in the study. Similarities and differences in art forms and art practices might be
observable with different models and communities. Another limitation is that while our participants are actively

16
The Prompt Artists Conference acronym ’XX, June 03–05, 2018, Woodstock, NY

involved in multiple communities, we did not ask participants about their experiences with other models or other
communities in depth.

8 CONCLUSION
In this paper, we have described a unique moment in time: Recent text-to-image models have given rise to an
exceptionally vibrant community of practice, complete with new ideas about what constitute notable outcomes
(e.g., new styles, prompts as art, and prompt templates as art). As the larger artistic and creator community adopts
these new models and forms of art, there are clear ways in which the tools can improve to better support desired
practices including: 1) helping to discover and create novel outputs, 2) providing methods to validate novelty,
and 3) elevating the notion of a prompt template into a standalone, first class, interactive object. In considering
design implications for these TTI models, our results also suggest the value in distinguishing between prompt
artists—those users who embrace the constraint of creating images using only an input prompt—and practitioners,
who may desire more fine-grained input and editing controls in comparison.

ACKNOWLEDGMENTS
REFERENCES
[1] [n.d.]. AICAN. https://www.aican.io/. (Accessed on 08/31/2022).
[2] [n.d.]. Imagen: Text-to-Image Diffusion Models. https://imagen.research.google/. (Accessed on 08/31/2022).
[3] [n.d.]. Midjourney. https://www.midjourney.com/home/. (Accessed on 09/01/2022).
[4] [n.d.]. Parti: Pathways Autoregressive Text-to-Image Model. https://parti.research.google/. (Accessed on 08/31/2022).
[5] [n.d.]. Stable Diffusion launch announcement — Stability.Ai. https://stability.ai/blog/stable-diffusion-announcement. (Accessed on
08/31/2022).
[6] Blaise Agüera y Arcas. 2017. Art in the Age of Machine Intelligence. Arts 6, 4 (2017). https://doi.org/10.3390/arts6040018
[7] Virginia Braun and Victoria Clarke. 2006. Using thematic analysis in psychology. Qualitative research in psychology 3, 2 (2006), 77–101.
[8] Andrew Brock, Jeff Donahue, and Karen Simonyan. 2018. Large Scale GAN Training for High Fidelity Natural Image Synthesis. CoRR
abs/1809.11096 (2018). arXiv:1809.11096 http://arxiv.org/abs/1809.11096
[9] Kieran Browne. 2022. Who (or What) Is an AI Artist? Leonardo 55, 2 (04 2022), 130–134. https://doi.org/10.1162/leon_a_02092
arXiv:https://direct.mit.edu/leon/article-pdf/55/2/130/2004755/leon_a_02092.pdf
[10] Daniel Buschek, Lukas Mecke, Florian Lehmann, and Hai Dang. 2021. Nine Potential Pitfalls when Designing Human-AI Co-Creative
Systems. Workshops at the International Conference on Intelligent User Interfaces (IUI) (2021).
[11] Eva Cetinic and James She. 2022. Understanding and Creating Art with AI: Review and Outlook. ACM Trans. Multimedia Comput.
Commun. Appl. 18, 2, Article 66 (Feb 2022), 22 pages. https://doi.org/10.1145/3475799
[12] Eugene Ch’ng. 2019. Art by Computing Machinery: Is Machine Art Acceptable in the Artworld? ACM Trans. Multimedia Comput.
Commun. Appl. 15, 2s, Article 59 (Jul 2019), 17 pages. https://doi.org/10.1145/3326338
[13] Christie’s. 2018. The first piece of AI-generated art to come to auction | Christie’s. https://www.christies.com/features/a-collaboration-
between-two-artists-one-human-one-a-machine-9332-1.aspx. (Accessed on 09/06/2022).
[14] John Joon Young Chung, Minsuk Chang, and Eytan Adar. 2021. Gestural Inputs as Control Interaction for Generative Human-AI
Co-Creation. Workshops at the International Conference on Intelligent User Interfaces (IUI) (2021).
[15] John Joon Young Chung, Shiqing He, and Eytan Adar. 2021. The Intersection of Users, Roles, Interactions, and Technologies in Creativity
Support Tools. In Designing Interactive Systems Conference 2021 (DIS ’21). Association for Computing Machinery, New York, NY, USA,
1817–1833. https://doi.org/10.1145/3461778.3462050
[16] Nicholas Davis, Chih-PIn Hsiao, Kunwar Yashraj Singh, Lisa Li, Sanat Moningi, and Brian Magerko. 2015. Drawing Apprentice: An
Enactive Co-Creative Agent for Artistic Collaboration. In Proceedings of the 2015 ACM SIGCHI Conference on Creativity and Cognition
(C&C ’15). Association for Computing Machinery, New York, NY, USA, 185–186. https://doi.org/10.1145/2757226.2764555
[17] Nicholas Davis, Chih-PIn Hsiao, Kunwar Yashraj Singh, Lisa Li, and Brian Magerko. 2016. Empirically Studying Participatory Sense-
Making in Abstract Drawing with a Co-Creative Cognitive Agent. In Proceedings of the 21st International Conference on Intelligent User
Interfaces (IUI ’16). Association for Computing Machinery, New York, NY, USA, 196–207. https://doi.org/10.1145/2856767.2856795
[18] Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. BERT: Pre-training of Deep Bidirectional Transformers
for Language Understanding. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational
Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers). Association for Computational Linguistics, Minneapolis,
Minnesota, 4171–4186. https://doi.org/10.18653/v1/N19-1423

17
Conference acronym ’XX, June 03–05, 2018, Woodstock, NY Chang et al.

[19] Prafulla Dhariwal and Alexander Nichol. 2021. Diffusion Models Beat GANs on Image Synthesis. In Advances in Neural Information
Processing Systems, M. Ranzato, A. Beygelzimer, Y. Dauphin, P.S. Liang, and J. Wortman Vaughan (Eds.), Vol. 34. Curran Associates, Inc.,
8780–8794. https://proceedings.neurips.cc/paper/2021/file/49ad23d1ec9fa4bd8d77d02681df5cfa-Paper.pdf
[20] Ahmed Elgammal, Bingchen Liu, Mohamed Elhoseiny, and Marian Mazzone. 2017. CAN: Creative Adversarial Networks, Generating
"Art" by Learning About Styles and Deviating from Style Norms. https://doi.org/10.48550/ARXIV.1706.07068
[21] Jonas Frich, Lindsay MacDonald Vermeulen, Christian Remy, Michael Mose Biskjaer, and Peter Dalsgaard. 2019. Mapping the Landscape
of Creativity Support Tools in HCI. In Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems (CHI ’19).
Association for Computing Machinery, New York, NY, USA, 1–18. https://doi.org/10.1145/3290605.3300619
[22] Leon A. Gatys, Alexander S. Ecker, and Matthias Bethge. 2015. A Neural Algorithm of Artistic Style. https://doi.org/10.48550/ARXIV.
1508.06576
[23] Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio.
2020. Generative Adversarial Networks. Commun. ACM 63, 11 (Oct 2020), 139–144. https://doi.org/10.1145/3422622
[24] Imke Grabe, Miguel González-Duque, Sebastian Risi, and Jichen Zhu. 2022. Towards a Framework for Human-AI Interaction Patterns in
Co-Creative GAN Applications. Workshops at the International Conference on Intelligent User Interfaces (IUI) (2022).
[25] Dejan Grba. 2019. Forensics of a molten crystal: challenges of archiving and representing contemporary generative art. ISSUE Annual
Art Journal: Erase 8, 3-15 (2019), 5.
[26] Dejan Grba. 2021. Brittle Opacity: Ambiguities of the Creative AI. In Proceedings of the xCoAx, 9th Conference on Computation,
Communication, Aesthetics & X Proceedings, xCoAx, Graz, Austria. 12–16.
[27] Dejan Grba. 2022. Deep Else: A Critical Framework for AI Art. Digital 2, 1 (2022), 1–32. https://doi.org/10.3390/digital2010001
[28] Aaron Hertzmann. 2018. Can Computers Create Art? Arts 7, 2 (2018). https://doi.org/10.3390/arts7020018
[29] Aaron Hertzmann. 2020. Computers Do Not Make Art, People Do. Commun. ACM 63, 5 (Apr 2020), 45–48. https://doi.org/10.1145/3347092
[30] Jonathan Ho, Chitwan Saharia, William Chan, David J Fleet, Mohammad Norouzi, and Tim Salimans. 2022. Cascaded Diffusion Models
for High Fidelity Image Generation. J. Mach. Learn. Res. 23 (2022), 47–1.
[31] Rania Hodhod and Brian Magerko. 2016. Closing the Cognitive Gap between Humans and Interactive Narrative Agents Using Shared
Mental Models. In Proceedings of the 21st International Conference on Intelligent User Interfaces (IUI ’16). Association for Computing
Machinery, New York, NY, USA, 135–146. https://doi.org/10.1145/2856767.2856774
[32] Cheng-Zhi Anna Huang, Hendrik Vincent Koops, Ed Newton-Rex, Monica Dinculescu, and Carrie J. Cai. 2020. AI Song Contest:
Human-AI Co-Creation in Songwriting. (2020). https://doi.org/10.48550/ARXIV.2010.05388
[33] Angel Hsing-Chi Hwang. 2022. Too Late to Be Creative? AI-Empowered Tools in Creative Processes. In Extended Abstracts of the 2022
CHI Conference on Human Factors in Computing Systems (CHI EA ’22). Association for Computing Machinery, New York, NY, USA,
Article 38, 9 pages. https://doi.org/10.1145/3491101.3503549
[34] Youngseung Jeon, Seungwan Jin, Patrick C. Shih, and Kyungsik Han. 2021. FashionQ: An AI-Driven Creativity Support Tool for
Facilitating Ideation in Fashion Design. In Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems (CHI ’21).
Association for Computing Machinery, New York, NY, USA, Article 576, 18 pages. https://doi.org/10.1145/3411764.3445093
[35] Yongcheng Jing, Yezhou Yang, Zunlei Feng, Jingwen Ye, Yizhou Yu, and Mingli Song. 2020. Neural Style Transfer: A Review. IEEE
Transactions on Visualization and Computer Graphics 26, 11 (2020), 3365–3385. https://doi.org/10.1109/TVCG.2019.2921336
[36] Pegah Karimi, Nicholas Davis, Mary Lou Maher, Kazjon Grace, and Lina Lee. 2019. Relating Cognitive Models of Design Creativity to
the Similarity of Sketches Generated by an AI Partner. In Proceedings of the 2019 on Creativity and Cognition (C&C ’19). Association for
Computing Machinery, New York, NY, USA, 259–270. https://doi.org/10.1145/3325480.3325488
[37] Tero Karras, Samuli Laine, and Timo Aila. 2019. A Style-Based Generator Architecture for Generative Adversarial Networks. In
Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).
[38] Ryan Louie, Andy Coenen, Cheng Zhi Huang, Michael Terry, and Carrie J. Cai. 2020. Novice-AI Music Co-Creation via AI-Steering Tools
for Deep Generative Models. In Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems (CHI ’20). Association
for Computing Machinery, New York, NY, USA, 1–13. https://doi.org/10.1145/3313831.3376739
[39] Elman Mansimov, Emilio Parisotto, Jimmy Lei Ba, and Ruslan Salakhutdinov. 2015. Generating Images from Captions with Attention.
https://doi.org/10.48550/ARXIV.1511.02793
[40] Marian Mazzone and Ahmed Elgammal. 2019. Art, Creativity, and the Potential of Artificial Intelligence. Arts 8, 1 (2019). https:
//doi.org/10.3390/arts8010026
[41] Jon McCormack, Toby Gifford, Patrick Hutchings, Maria Teresa Llano Rodriguez, Matthew Yee-King, and Mark d’Inverno. 2019. In a
Silent Way: Communication Between AI and Improvising Musicians Beyond Sound. In Proceedings of the 2019 CHI Conference on Human
Factors in Computing Systems (CHI ’19). Association for Computing Machinery, New York, NY, USA, 1–11. https://doi.org/10.1145/
3290605.3300268
[42] Matt McFarland. 2016. Google’s psychedelic ‘paint brush’ raises the oldest question in art - The Washington Post. https://www.
washingtonpost.com/news/innovations/wp/2016/03/10/googles-psychedelic-paint-brush-raises-the-oldest-question-in-art/. (Accessed
on 09/14/2022).

18
The Prompt Artists Conference acronym ’XX, June 03–05, 2018, Woodstock, NY

[43] Matthew B Miles and A Michael Huberman. 1984. Drawing valid meaning from qualitative data: Toward a shared craft. Educational
researcher 13, 5 (1984), 20–30.
[44] Alexander Mordvintsev, Christopher Olah, and Mike Tyka. 2015. Inceptionism: Going Deeper into Neural Networks. https://research.
googleblog.com/2015/06/inceptionism-going-deeper-into-neural.html
[45] Michael Muller, Justin D Weisz, and Werner Geyer. 2020. Mixed Initiative Generative AI Interfaces: An Analytic Framework for Generative
AI Applications. In Proceedings of the Workshop The Future of Co-Creative Systems-A Workshop on Human-Computer Co-Creativity of the
11th International Conference on Computational Creativity (ICCC 2020).
[46] Changhoon Oh, Jungwoo Song, Jinhan Choi, Seonghyeon Kim, Sungwoo Lee, and Bongwon Suh. 2018. I Lead, You Help but Only
with Enough Details: Understanding User Experience of Co-Creation with Artificial Intelligence. In Proceedings of the 2018 CHI
Conference on Human Factors in Computing Systems (CHI ’18). Association for Computing Machinery, New York, NY, USA, 1–13.
https://doi.org/10.1145/3173574.3174223
[47] OpenAI. [n.d.]. DALL·E 2. https://openai.com/dall-e-2/. (Accessed on 08/31/2022).
[48] OpenAI. [n.d.]. DALL·E: Creating Images from Text. https://openai.com/blog/dall-e/. (Accessed on 08/31/2022).
[49] Allison Perrone and Justin Edwards. 2019. Chatbots as Unwitting Actors. In Proceedings of the 1st International Conference on
Conversational User Interfaces (CUI ’19). Association for Computing Machinery, New York, NY, USA, Article 2, 2 pages. https:
//doi.org/10.1145/3342775.3342799
[50] Brian Quanz, Wei Sun, Ajay Deshpande, Dhruv Shah, and Jae-eun Park. 2020. Machine learning based co-creative design framework.
arXiv preprint arXiv:2001.08791 (2020).
[51] Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, Peter J Liu, et al. 2020.
Exploring the limits of transfer learning with a unified text-to-text transformer. J. Mach. Learn. Res. 21, 140 (2020), 1–67.
[52] Aditya Ramesh, Prafulla Dhariwal, Alex Nichol, Casey Chu, and Mark Chen. 2022. Hierarchical Text-Conditional Image Generation
with CLIP Latents. https://doi.org/10.48550/ARXIV.2204.06125
[53] Aditya Ramesh, Mikhail Pavlov, Gabriel Goh, Scott Gray, Chelsea Voss, Alec Radford, Mark Chen, and Ilya Sutskever. 2021. Zero-Shot
Text-to-Image Generation. In Proceedings of the 38th International Conference on Machine Learning (Proceedings of Machine Learning
Research), Marina Meila and Tong Zhang (Eds.), Vol. 139. PMLR, 8821–8831. https://proceedings.mlr.press/v139/ramesh21a.html
[54] Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Björn Ommer. 2022. High-Resolution Image Synthesis With
Latent Diffusion Models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 10684–10695.
[55] Chitwan Saharia, William Chan, Saurabh Saxena, Lala Li, Jay Whang, Emily Denton, Seyed Kamyar Seyed Ghasemipour, Burcu Karagol
Ayan, S. Sara Mahdavi, Rapha Gontijo Lopes, Tim Salimans, Jonathan Ho, David J Fleet, and Mohammad Norouzi. 2022. Photorealistic
Text-to-Image Diffusion Models with Deep Language Understanding. https://doi.org/10.48550/ARXIV.2205.11487
[56] Othman Sbai, Mohamed Elhoseiny, Antoine Bordes, Yann LeCun, and Camille Couprie. 2018. DesIGN: Design Inspiration from Generative
Networks. In Proceedings of the European Conference on Computer Vision (ECCV) Workshops.
[57] Yang Shi, Nan Cao, Xiaojuan Ma, Siji Chen, and Pei Liu. 2020. EmoG: Supporting the Sketching of Emotional Expressions for
Storyboarding. In Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems (CHI ’20). Association for Computing
Machinery, New York, NY, USA, 1–12. https://doi.org/10.1145/3313831.3376520
[58] Angie Spoto and Natalia Oleynik. [n.d.]. Library of Mixed-Initiative Creative Interfaces. http://mici.codingconduct.cc/aboutmicis/.
(Accessed on 08/31/2022).
[59] Georgios N Yannakakis, Antonios Liapis, and Constantine Alexopoulos. 2014. Mixed-initiative co-creativity. (2014).
[60] Jiahui Yu, Yuanzhong Xu, Jing Yu Koh, Thang Luong, Gunjan Baid, Zirui Wang, Vijay Vasudevan, Alexander Ku, Yinfei Yang, Burcu Karagol
Ayan, Ben Hutchinson, Wei Han, Zarana Parekh, Xin Li, Han Zhang, Jason Baldridge, and Yonghui Wu. 2022. Scaling Autoregressive
Models for Content-Rich Text-to-Image Generation. https://doi.org/10.48550/ARXIV.2206.10789
[61] Nanxuan Zhao, Nam Wook Kim, Laura Mariah Herman, Hanspeter Pfister, Rynson W.H. Lau, Jose Echevarria, and Zoya Bylinskii.
2020. ICONATE: Automatic Compound Icon Generation and Ideation. In Proceedings of the 2020 CHI Conference on Human Factors in
Computing Systems (CHI ’20). Association for Computing Machinery, New York, NY, USA, 1–13. https://doi.org/10.1145/3313831.3376618
[62] Jun-Yan Zhu, Taesung Park, Phillip Isola, and Alexei A. Efros. 2017. Unpaired Image-To-Image Translation Using Cycle-Consistent
Adversarial Networks. In Proceedings of the IEEE International Conference on Computer Vision (ICCV).

19