Open Debate

AI Fantasy and (Anti)

Co-Creative Machines
On Singularity Dadaism,
Syntropic Counterpoints
and Design

Predrag K. Nikolić Giacomo Bertin

College For Creative Studies Università degli Studi di Padova
[email protected] [email protected]
ORCID 0000-0003-1409-088X ORCID 0009-0006-6294-1940

The recent step forward in questioning Artificial Intelligence Keywords

(AI) creativity, enticed by deep learning models such as GPT Generative storytelling
and Stable Diffusion, instilled into the design community of Natural language processing
educators and practitioners the idea of raising new creative Artificial Intelligence (AI)
practices worth discussion and analysis. Hence, Natural creativity
Language Processing (NLP) and computer vision are filled Human-AI interaction
with encounters of interesting creative responses by robots Computer vision
and are becoming a challenging creative medium for artists
and designers (Hemanth, 2023). In this paper, we will pres-
ent several experimental projects in the arts and design that
we created to raise and discuss various questions related
to the future development of AI imagination (Nikolić, 2019),
the liberation of machine creativity (Nikolić & Liu, 2021) and
our conceptual approach, which challenges AI aesthetics
and machine abstraction (Nikolić & Bertin, 2023). A key
novelty in the proposed creative practice is the design of
AI clones under novel domains of knowledge to use as a
creative medium for machine-supervised content creation
(Nikolic, 2021).

Copyright © 2023
diid disegno industriale
industrial design
58 Predrag K. Nikolić, Giacomo Bertin CC BY-NC-SA
 Introduction

Artificial Intelligence (AI) is integrated into our lives so subtly that we

are not even aware of it in many cases. It is everywhere, in our cars,
houses, pockets, public spaces, and workplaces, and it continues to
penetrate every segment of human society (Hemanth, 2023). In the
beginning, it was just a tool for robust calculations and analysis, but
throughout the years AI has become a substantial entity in our daily
interactions and, most recently, part of new tendencies in creative
practices (Oliveira, 2022).
This paper will present experimental works by Predrag K.
Nikolić as part of his research project Syntropic Counterpoints from
2016-2023. We will use conceptual and creative intentions behind
those works (Nikolić, 2019) to raise several issues regarding the
application of AI in art and design with a focus on autonomous gener-
ative storytelling (Nikolić & Liu, 2021) and computer vision (Nikolić &
Bertin, 2023). Accordingly, we will discuss new evaluation criteria for
such an act of creation, conceived in the proposed idea of Singularity
Dadaism (Nikolic, 2021), as well as the ethical and mutual trust-related
consequences of such an approach (Nikolić & Tomari, 2021). We start
with the following question: What level of trust can — and should —
we place in AI systems we are employing to support us in what we are
doing and what we think?

AI Creativity in the Triangle of Trust

When discussing AI and the human values necessary to embed into

its reasoning, we usually refer to morals and ethics as significant influ-
ences on our decision-making. Nevertheless, we must also encoun-
ter a cultural dimension, as typical behaviour in one ethnic group or
region could be completely different elsewhere. This complicates our
effort to integrate morality in AI systems, as we cannot apply univer-
sally accepted ethical beliefs (Nikolić & Tomari, 2021).
Our notion of ethics is influenced by our feelings, ideas of
rights, and the social and cultural environment in which we grow, live,
and share similar values. Accordingly, if artificial agents cannot feel the
consequences of their decisions on an emotional level like a human,
even if we employ an acceptable universal ethical and moral system,
the result can end in failure. So, can we trust AI, and is it a thing to be
trusted? What is trust, and how does it reflect on our behaviour?
“Trust” is helping humans to build their relationships and keep
them together. It is a link between peers in which the trusting party
believes in any promises made by the trusted party (Bryson, 2018).
In most cases, we can negotiate and navigate trust issues subcon-
sciously. However, our living environment is expanding toward new
creative practices and interactions with intelligent artificial entities,
such as the prospect of integrating robots more closely into everyday
life. Therefore, we must question the nature and value of trust under
novel socio-technological circumstances. AI is a system developed
to compute actions or knowledge from a data set. As such, does it
have the credibility to be trusted? Accordingly, only other software
developed by using the same techniques can be a peer with AI. There
is a widely instilled opinion that no one can trust AI, and only humans

diid No. 80 — 2023

59 AI Fantasy and (Anti) Co-Creative Machines Doi: 10.30682/diid8023e
can hold that system and take responsibility. The logical question to
raise is, how can we authorise AI to make decisions in our place if
we do not trust them? Can we develop an AI entity that will follow the
ethical and moral principles we will trust?
It is a challenge we must include in future research related
to Human-AI interaction. Several authors addressed interpersonal
trust as most relevant to human-machine interaction instead of insti-
tutional trust. They use a multidisciplinary approach that includes
philosophy, law, and neuroscience to explore the issue of trust in
Human-AI interactions. They expressed confidence in achieving
interpersonal trust between humans and robots but also underlined
the potential damage that could arise from facilitating it in such inter-
actions (Kirkpatric, Hahn & Haufler, 2017). If we accept the thesis
that humans and robots could trust each other, we immediately face
another relevant issue: Should they distrust each other? (Lin, Abney
& Jenkins, 2017)
Humans communicate in a very particular way, often using
all manner of deception, including sarcasm, analogies, innuendo,
and other oblique speech forms. This must therefore be included in
Human-AI interaction as a moral necessity. Social robots will need
the ability to detect and evaluate such deceptive speech, avoid being
manipulated and cause unintended damage (Isaac & Bridewell,
2017). When discussing the future of human-robot interaction, we
need to consider the possibility that robots will become so advanced
that they can develop interpersonal relationships and trust, including
features like deception.

Syntropic Counterpoints: Robosophy Philosophy

Predrag Nikolić’s installation Syntropic Counterpoints: Robosophy

Philosophy is conceptualised as AI robot-robot-conducted philo-
sophical discussions. For that purpose, we created two AI Philos-
opher Clones, Aristotle and Nietzsche (Nikolic & Yang, 2020). We
presented them with discourse related to topics such as morality,
ethics, and aesthetics. We intended to expose the AI system to their
interpretation of highly contemplative abstract knowledge related to
human virtues, problem definition, trust, and decision-making, which
we found significant for the act of creation. Through this genuine
philosophical robot-robot interaction that happened in the total
absence of people, we tried to question ongoing cultural and social
changes, which are the results of interactions between people and
technology (Nikolić, 2019).
Moreover, the potential consequences of giving the power
of the machine to interpret human knowledge and aesthetical values
are essential for creativity and human society’s well-being. We have
had the intention to trigger an epical discussion between Aristot-
le’s Ethical Robot (Magnanimous) and Nietzche’s Superman Robot
(Übermensch) Fig. 1. Magnanimous follows noble causes based on
virtue and traditional spiritual values, and Übermensch strives to
create new values and goals for humanity.

60 Predrag K. Nikolić, Giacomo Bertin

 Fig. 1
Aristotle’s Ethical Robot
and Nietzche’s Super-
man Robot. Ph. Predrag
K. Nikolić © Predrag K.
Nikolić.

These robots are not robots. They are pseudo-robots, just like their
discussion which is a pseudo-philosophical discussion. Neverthe-
less, we need to address the possibility of establishing new values
in the man-machine hybrid society of the future and instil the next
generation of human-AI interactions and creative implications with
that transformation. Especially if those two entities (human and AI
systems) start to separate from each other and machines become
autonomous, anti-social, and anti-co-creative in an unexpected way.
Shall we trust them, and to what extent will they be capable of inter-
preting the emotions and human values we involve in the creative
design process?

Singularity and Metaphysics of the Machines

At the same time, we tried to investigate the value of words as part

of the fundamental data we are using to transfer knowledge and
interact with AI entities. In our approach, we used the analogy with
Tristan Tzara’s statement: “My words are not mine. My words are
everybody else’s words: I mix them very nicely”. This demonstrates
the Dadaistic aesthetic of the absurd with subliminal messages
given to the audience to interpret freely. Similarly, our AI Philosopher
Clones are mixing the words of Aristotle and Nietzsche, and those
words are not their words. They are cold calculations, algorithms, or
a still unknown level of abstraction achieved by machines.
Suppose grammar is the “metaphysics of the people”, as
Nietzsche claimed (Guthrie, 2016). In that case, discussions in the
installation are “metaphysics of the machines”, done by anti-social
robots. They prefer to keep humans from this process, interacting
with each other and acting upon their logic to achieve results known
only to them. Hence, does that mean that the future shaping of
design trends, innovation, and aesthetical criteria will end up in the
hands of autonomous anti-social and anti-co-creative AI agents we
should not trust? Does that mean the new hybrid society aesthetic
will be a poor interpretation of human needs and desires?

diid No. 80 — 2023

61 AI Fantasy and (Anti) Co-Creative Machines Doi: 10.30682/diid8023e
 Syntropic Counterpoints: Metaphysics of the Machines

In Predrag Nikolić’s installation Syntropic Counterpoints: Metaphys-

ics of The Machines, we explore the AI aesthetics phenomena and
challenge machine abstraction. We draw parallels between Tech-
nological Singularity (Shananah, 2015) and Dadaism to join them
into Singularity Dadaism and propose novel creative practices run
by liberated anti-co-creative machines. We exposed AI Philosopher
Clones to some of the fundamental philosophical questions about
the relationship of thought to being and metaphysics. For that pur-
pose, we created four AI Philosopher Clones: Aristotle, Nietzsche,
Sun Tzu, and Machiavelli, who were trained with some of the crucial
philosophical discourses these philosophers left to humankind. We
are triggering their discussion with a set of eternal questions such
as “Why is there something rather than nothing?”, “Is war moral and
ethical, and can it ever be justifiable?” or “What is good and what is
evil?” and then leave the discussion entirely in the hands of the AI
clones to develop it further, make their questions based on the gen-
erated content and tackle other topics such as progress, law, moral-
ity and all the other fine qualities that various intelligent men have
discussed in so many books throughout the history of humankind.
The AI Philosopher Clones follow their patterns and use a creative
vocabulary made of the world, grammar, and letters to offer one more
alternative reality. It is “beyond” a world made of calculations, algo-
rithms, predictions, probabilities, and machine-made choices that we
expect to sort our life’s mysteries, give answers to eternal questions
humankind can ask, and make decisions for us (Nikolić & Liu, 2021).
With the installation Metaphysics of The Machines, we are comment-
ing on such attitudes using irony and scepticism and questioning the
human capacity to understand creations beyond logic and pragma-
tism, the worlds beyond human perception Fig. 2.
Fig. 2
Interactive Installation
Metaphysics of The
Machines, antagonistic
relationship between the
AI Philosophers and visi-
tors. Ph. Predrag K. Nikolić
© Predrag K. Nikolić.

Our AI Philosopher Clones are in complete control in conducting and

materialising their philosophical discussion into alternative reality
projected on the wall. Our focus is on words, how AI clones are
putting them together, and the vocal organs humans use to formulate

62 Predrag K. Nikolić, Giacomo Bertin

their thoughts and transfer them into speech. The result is unpredict-
able, unknown until it happens, and it questions the whole concept
of knowledge. It is content generated from the written corpus of
highly intelligent people and interpreted by the machines to which
we gave control to imagine worlds beyond the world for us.

AI Fantasy in AI.R Taletorium

AI.R Taletorium is an AI-based storytelling system that offers children

with special needs the possibility to experience autonomous AI fairy
tale storytelling. The system can, but not necessarily does, involve
kids in the creative process. Our focus was on the generation by AI
of one of the most demanding narratives when it comes to human
fantasy, fairy tales (Nikolić & Bertin, 2023).
Furthermore, in this project we prototyped the concept
of Artificial Intelligence Reality (AI.R) as a novel reality paradigm,
designed with robot creativity and AI processed data collected via
sensors and cameras from the environment (Nikolić, 2019). Hence,
in addition to textual datasets and visual data analyses, we use
users’ facial features as inputs for AI to create authentic fairy tales.
We used the human-machine (anti) co-creation concept, AI facial
recognition, GPT-3 neural network, AI RNN sketch recognition, and
AI visualisation to design a generative storytelling model. The num-
ber of users involved in the co-creation of fairy tales is unlimited, but
the system decides what to add or not in the storyline. An AI agent
collects users’ sketched objects and characters and integrates them
into the storyline. The system offers a novel creative experience with
machines and humans joined together (under machine rules) in the
act of creation.

Designing the (Anti) Co-Creating Machines

Social Robotics encompasses the meaning of “social”. It represents

two or more entities within the same context (Duffy et al., 1999) and
is sociable, related to the willingness to talk and interact with others.
Accordingly, social robots such as co-creative robots were explicitly
developed for the interaction between humans and machines, to
support processes through a human-like interaction. Nevertheless,
the absence of shared ethical and moral criteria creates obstacles
that are hard to overcome, especially when interacting with a sensi-
tive population. So, more Human-AI social interaction or anti-social
and (anti) co-creative machines can quickly become a new reality we
will be forced to deal with (Tulli et al., 2019).

diid No. 80 — 2023

63 AI Fantasy and (Anti) Co-Creative Machines Doi: 10.30682/diid8023e
 AI.R Taletorium (Anti) Co-Creative Machine

AI.R Taletorium is a character-centric multi-modal AI fairy tale-tell-

ing system that aims to connect participants within a unified fairy
tale creation process using AI (Nikolić & Bertin, 2023). The system
consists of:
• A generator of characters based on users’ facial properties
• AI voice storyteller
• AI scene generator
• Interface for real-time user content contribution.
To achieve participant immersion in the AI-generated fantasy,
we used the paradigm of avatars or players’ virtual characters in vir-
tual (gaming) worlds (Christy & Fox, 2016) and feelings of self-pres-
ence in the AI-created virtual world.
To develop our co-creative model, we used a combination
of the following: FaceNet (Schroff, Kalenichenko & Philbin, 2015)
for facial landmark recognition, Mxnet (Chen et al., 2015) for facial
attribute extraction, AgeNet (Liu et al., 2015), a pre-trained Caffe
model used for age detection, and GenderNet: Caffe model for
gender detection. While creating users’ avatars for later placement
in the story as characters, we combine multiple facial characteristics
collected by the webcam to extract the attributes using the main
algorithm with pre-trained FaceNet, MxNet, AgeNet, and GenderNet.
We then combined the basic version of the facial prototype
with the AI story generation model to take facial pictures, map them
into corresponding characters, and subsequently generate a sto-
ryline and content (Nikolić & Bertin, 2023). Based on the plan-and-
write NLP story generation model, we trained two sub-models to plan
and generate a story (Yao et al., 2018). As for training, we collected
890 fairy tales and trained our two sub-models with them. We applied
a modified RAKE algorithm to generate the keywords as the storyline
for each story example (Rose et al., 2010).
After generating the story, our AI system generates the
described scenes in generated fairytales, starting with a 3D object
from a text description using propositions from similar visualisation
issues (Liu et al., 2015). We regularise mesh generation without Lap-
lacian loss and optimise two nets instead of normal maps, positions,
and texture. We represent the 3D object as a set of vertices: V ∈ R3,
a 2D texture T, and faces. In our pipeline, we do not directly optimise
the positions of the vertices. Instead, we trained a Neural Mesh Flow
(NMF) (Christy & Fox, 2016) network that acts as a deformer. The tex-
ture is generated by a SirenNet (SN) (Sitzmann et al., 2020). We found
that using this architecture instead of directly optimising a 2D image
improved texture quality and reduced the time for convergence Fig. 3.

64 Predrag K. Nikolić, Giacomo Bertin

 Fig. 3
Overview of optimization
pipeline: we optimise a
Neural Mesh Flow Net
(Gupta & Chandraker,
2020) and a SirenNet
(Sitzmann et al., 2020) to
learn a deformation and
a texture to maximise the
similarity between the
encoding of a prompt text
and multiple views with
the random background.

With Pytorch3d differentiable render, we generate a batch of images

from multiple view angles and distances Ii = R(V, T, θi, ϕi, di). We use
CLIP (Radford et al., 2021) to encode images in the same embedded
space of the text prompt Fig. 4.
Fig. 4
Sample of generated
images. A knight, a ghost,
and a Halloween pumpkin
near a dead tree. A Gothic
Castle on foggy hills on
a night.

Finally, we compute the loss as the average cosine similarity

between the encoded images and the text. To encourage a better
frontal view representation, we sample azimuth ϕ from a beta distri-
bution with α = 1.0, β = 5.0 within a range of [180◦, 180◦]. In contrast,
elevation θ is uniformly sampled between [10◦, 45◦] and distance d
in [4.2, 5.7]. To force the net to focus on the shape, the mesh centre
is translated from a vector ϵ sampled from a normal distribution with
σ = 0.5. The weights of NMF Net are adjourned three times more
frequently than the texture. The background is generated by ran-
domly merging a chessboard with a casual number of squares and

diid No. 80 — 2023

65 AI Fantasy and (Anti) Co-Creative Machines Doi: 10.30682/diid8023e
a random noise, then− augmented with Gaussian blur and random
− graph, the system generates a loss
desaturation.Given the scene
that represents the discrepancy between the actual configuration
and the target scene graph, then chooses the variables to optimise
according to the objects’ relations (if an object is “on” a plane, only
two parameters are needed). Finally, it finds the optimal translation
and rotation for visualising objects. Finally, positions can be refined
using CLIP as before Fig. 5.
Fig. 5
Scene Generation. Find
the configuration that min-
imises a loss representing
the agreement between
this configuration and the
scene graph.

Virtual Skin (ANTI) Co-Creative Machine

The project Virtual Skin aims to explore the possibilities of using

AI to generate 3D virtual materials and for their potential to arouse
sufficient sensory and emotional stimuli. We collected and merged
multi-sensory inputs to design AI-generated 3D materials for virtual
and mixed-reality products and environments.
Starting from text descriptions, we used DALL-E to generate
the image of a sphere texturized according to the initial prompt, and
then we extracted the corresponding texture. The procedure to fit
the bump map and the texture uses the Stable Diffusion model to
progressively refine the starting template mesh, which in our case,
is a sphere.
A neural network generates an RGB texture and predicts
the magnitude of the movement along the normal direction for each
vertex. To compute the displacement, we encode the position of the
vertex using Multiresolution Hash Encoding (MHE). We parsed the
features and obtained to a SirenNet, an MLP (MultiLayer Percep-
tron), where are the vertices normal and the maximum displace-
ment. The texture is computed similarly mapping the positions of
the pixels to the rang . This initial transformation ensures periodical
textures. Using a sinusoidal activation function makes the training
shorter and more stable for this task. During the generation, we opti-
mise the material gloss level to reduce artefacts, like light reflection
included in the texture Fig. 6.

66 Predrag K. Nikolić, Giacomo Bertin

 Fig. 6
AI Synesthetic Virtual
Materials Generation
Procedure.

For each training step, we start with generating the texture and the
displacement along the normal for each vertex. Then we produce
a refined version of each render using the Stable Diffusion pipe-
line conditioned on the text prompt. Finally, we compute the MSE
(Mean Square Error) between the main render and the target image.
Then, we compare the renders from the cameras sampled uniformly
around the sphere and their refined version using a perceptual loss
based on a pre-trained VGG19 net.

Conclusion

In this paper, we presented an extensive corpus of experiments and

the use of AI on different topics and zones of interest, such as trust,
imagination, aesthetics, ethics, and morals. The landscape of oppor-
tunities and concerns is vast, and every paradigm in that field repre-
sents a challenge for future generations of designers and engineers
involved in this development.
AI as a storyteller, philosopher, or with synesthetic capabil-
ities, are just some examples of AI utilisation toward new creative
practices and realms. These custom-trained neural networks show
the limitations of AI technology regarding creativity, generative con-
tent, and placing the content in a particular context.
Without moving forward in the direction that will improve
machine reasoning and decision-making beyond algorithms and cold
calculations, its creative potential will remain limited and mostly rep-
licative rather than inspiring or innovative in its application in creative
practices such as fine arts, design, music, literature, or performing

diid No. 80 — 2023

67 AI Fantasy and (Anti) Co-Creative Machines Doi: 10.30682/diid8023e
arts. Nonetheless, developing such potential could lead to a total Predrag K. Nikolić
absence of human creative involvement, emotional triggers, and He is Chair and Full Profes-
sor of the Graduate User
sensory stimuli essential to understand and fulfil human needs with Experience (UX) Design
our creative production in the future. Program at the College for
New procedures for generating 3D meshes from text (such Creative Studies, Detroit,
United States. He is an
as Magic3D) still require considerable computational resources experimental designer,
and long execution times. A possible solution is the development of Editor-in-chief for EAI
2D diffusion models to generate multiple views of the same object. Endorsed Transactions
on Creative Technologies
This could help create synthetic datasets of objects to train 3D Journal, European Alliance
scattering models. for Innovation (EAI) Fellow
and founder of AI.R Lab.

Giacomo Bertin
He is a Physics of Data stu-
dent at the Università degli
Studi di Padova who holds a
Bachelor’s Degree in Phys-
ics with a thesis in compu-
tational methods for drug
discovery. He is a machine
learning engineer, AI devel-
oper, and co-founder of AI.R
Lab, focused on computer
vision and natural language
processing.

68 Predrag K. Nikolić, Giacomo Bertin

References

Bryson, J. (2018, Novem- Isaac, A. M., & Bridewell, W. Nikolic, P. K. (2021). Syn- Radford, A., Wook Kim, J.,
ber 14). No One Should (2017). White Lies on Silver tropic Counterpoints: Sin- Hallacy, C., Ramesh, A.,
Trust Artificial Intelli- Tongues. Why robots need gularity Dadaism as Novel Goh, G., Agarwal, S., Sastry,
gence. Our Word. https:// to deceive (and how). In P. AI Creative Practice Para- G., Askell, A., Mish- kin, P.,
ourworld.unu.edu/en/ Lin, K. Abney, & R. Jenkins digm. In K. Arai, S. Kapoor, Clark, J., Krueger, G., & Sut-
no-one-should-trust-artifi- (Eds.), Robot Ethics 2.0: & R. Bhatia (Eds.), Proceed- skever, I.. (2021). Learning
cial-intelligence From Autonomous Cars ings of the Future Technol- Transferable Visual Models
to Artificial Intelligence ogies Conference (FTC) from Natural Language
Chen, T., Li, M., Li, Y., Lin, (pp.157-172). Oxford Uni- 2020, Vol. 1. FTC 2020. Supervision. ArXiv. https://
M., Wang, N., Wang, M., versity Press. Advances in Intelligent doi.org/10.48550/arXiv.
Xiao, T., Xu, B., Zhang, Systems and Computing, 2103.00020
C., & Zhang, Z.. (2015). Kirkpatrick, J., Hahn, E. N., Vol 1288. Springer. https://
Mxnet: A flexible and & Haufler, A. J. (2017). Trust doi.org/10.1007/978-3-030- Rose, S., Engel, D., Cramer,
efficient machine learning and Human-Robot Inter- 63128-4_23 N., & Cowley, W. (2010).
library heterogeneous actions. In P. Lin, K. Abney, Automatic keyword
distributed systems. ArXiv. & R. Jenkins (Eds.), Robot Nikolić, P. K., & Liu, R. extraction from individual
https://doi.org/10.48550/ Ethics 2.0: From Auton- (2021). Metaphysics of documents. In M. W. Berry,
arXiv.1512.01274 omous Cars to Artificial The Machines: From & J. Kogan (Eds.), Text
Intelligence (pp. 142-156). Human-Robot-Robot Inter- Mining: Applications and
Christy, K. R., & Fox, J. Oxford University Press. action to AI Philosophers Theory, John Wiley & Sons,
(2016). Transportability and Abstraction. Artnodes, Hoboken (pp. 1-20). https://
Presence as Predictors of Lin, P., Abney, K., & Jenkins, 28, 1-10. UOC. https://doi. doi.org/10.1002/97804
Avatar Identification Within R. (Eds.). (2017). Robot Eth- org/10.7238/artnodes. 70689646.ch1
Narrative Video Games. ics 2.0: From Autonomous v0i28.385735
Cyberpsychology, Behavior, Cars to Artificial Intelli- Schroff, F., Kalenichenko,
and Social Networking, gence. Oxford University Nikolić, P. K., & Bertin, G. D., & Philbin, J. (2015).
19(4), 283-287. 10.1089/ Press. (2023). AI.R Taletorium: Facenet: A unified embed-
cyber.2015.0474 Artificial Intelligence 1001 ding for face recognition
Liu, X., Li, S., Kan, M., Cyber Nights. Artnodes, and clustering. 2015 IEEE
Duffy, B. R., Rooney, C., Zhang, J., Wu, S., Liu, 31. UOC. https://doi. Conference on Computer
O’Hare, G. M., & O’Dono- W., Han, H., Shan, S., & org/10.7238/artnodes. Vision and Pattern Recog-
ghue, R. (1999). What is a Chen, X. (2015). Agenet: v0i31.402821 nition (CVPR), Boston, MA
social robot?. In 10th Irish Deeply learned regressor (pp. 815-823). 10.1109/
Conference on Artificial and classifier for robust Nikolić, P. K., & Razali CVPR.2015.7298682
Intelligence & Cognitive apparent age estimation. Md Tomari, M. (2021).
Science, University College Proceedings of the IEEE Robot-Robot Interaction, Shanahan, M. (2015). The
Cork, Ireland, 1-3 Septem- International Conference Toward New Conversational Technological Singularity.
ber, 1999. on Computer Vision (ICCV) Artificial Intelligence Aes- MIT Press.
Workshops (pp. 16-24). thetic. 10th International
Gupta, K., & Chandraker, Conference on Digital and Sitzmann, V., Martel, J. N.
M. (2020). Neural Mesh Nikolić, P. K. (2019). Intelli- Interactive Arts (ARTECH P., Bergman, A., Lindell, D.
Flow: 3d Manifold Mesh gent Playful Environments 2021), October 13-15, 2021, B., & Wetzstein, G. (2020).
Generation via Diffeo- in New Urban Social Land- Aveiro, Portugal, Portugal Implicit neural rep-
morphic Flows”. ArXiv. scape. In H. Santos, G. Viale (pp. 1-9). ACM. https://doi. resentations with periodic
https://doi.org/10.48550/ Pereira, M. Budde, S. Lopes, org/10.1145/3483529 activation functions. ArXiv.
arXiv.2007.10973 & P. Nikolić (Eds.), Science .3483659 https://doi. org/10.48550/
and Technologies for Smart arXiv.2006.09661
Guthrie, B. (2016). Dada, Cities. International Summit Oliveira, A. (2022). Future
Nietzsche, and the Art Smart City 360° (pp. 36-45). Imaginings in Art and Arti- Tulli, S., Ambrossio, D. A.,
of Madness. https:// Springer. ficial Intelligence. Journal Najjar, A., & Lera, F. J. R.
blakeguthrie.wordpress. of Aesthetics and Phenom- (2019). Great Expectations
com/2016/11/04/dada- Nikolic, P. K., & Yang, H. enology, 9(2), 209-225. & Aborted Business Initia-
nietzsche-and-the-art-of- (2020). Artificial intelligence https://www.tandfonline. tives: The Paradox of Social
madness/ clone generated content com/doi/full/10.1080/2053 Robot Between Research
toward robot creativity 9320.2022.2150467 and Industry. BNAIC/BENE-
Hemanth, D. J. (2023). and machine mindfulness. LEARN.
Smart Applications with Mobile Networks and Appli-
Advanced Machine Learn- cations, 25, 1504-1513. Yao, L., Nanyun, P., Ralph,
ing and Human-Centred W., Kevin, K., Dongyan, Z.,
Problem Design. Springer & Rui, Y. (2018). Plan-And-
International Publishing. Write: Towards Better Auto-
matic Storytelling. ArXiv.
https://doi.org/10.48550/
arXiv.1811.05701

diid No. 80 — 2023

69 AI Fantasy and (Anti) Co-Creative Machines Doi: 10.30682/diid8023e