AI Narrative : Virtual Being interaction as a narrative device.

I had a query about whether can help me to create a character that will, on interaction, decide the fate of a player. So, for example, on meeting the mystic character, depending on whether the player pleases or aggravates her, she will decide if they go to heaven or go to hell (return to the start of the world again). 
Guy Gadney kindly advised that it is possible to create a single ‘scene’ that directs to two sub-plots. A condition could be put on each subplot. So for example the heaven route would lead to the experience end while the other returns the player to the start (in a non-linear way). 
I’m not entirely sure how this will work in practice but I’m thinking that it may be possible that interaction with mystic can be positive or negative depending on her AI responses (input from her back story). So for example, if she doesn’t like swearing then if the player swears at her she could send you on the track to hell.. back to the start. Whereas if you were to flatter her, she might warm to you and send you on to heaven (unless you are being insincere of course!). 

Lunarium: phase 3 Project management

I’m continuing with Lunarium project management using Trello. This can be accessed by anyone who is interested (it also details the full progress of the project).

Phase 3 will build from optimisation in unity and world design to now focus on Narrative. This will include character, set and prop design as well as AI.
So, I will be introducing Virtual Beings that will be placed within a carefully constructed virtual world (designed for cinematic narrative). 
Lunarium: Phase 3 Development 

AI Narrative : Narrative mapping (General/VR/AI)

In this post I’m going to further flesh out the relationship between world structure and narrative structure, look at the various models and apply these to my current projects. Ultimately this will lead to focusing on AI Narrative which will firm the bases of practical experimentation over the next few weeks, using the new app.

Many stories are written using the three act narrative model as it provides a strong narrative arc which emotionally engages the viewer. It is one which is very familiar and can be found in countless stories.

So for example, hero sets out on their quest in their first act, only to meet a character who leads them to experiencing the 2nd act, that which sets up the challenge they must overcome, which directs them down a line to overcoming this into the 3rd act of resolution. End.

In VR, the player is immersed in a world where they can seemingly go anywhere. They do not necessarily follow a linear track unless the world around them is designed in a way that gives them very little choice but to do so.

Players interaction with the environment or characters in the scene, are pivotal to where players could go next and so are a key narrative device.

My thoughts are that each act could be structured as three levels (3 scenes in unity). Within these worlds there could be a multitude of characters, environments and endings (When I think about this I can’t help but chuckle at the many days and weeks spent reading the fantasy fighting novels as a teenager, and particularly the many fingered and thumbed marking of pages as if doing some kind of hand yoga).

This structure would lead to a high level of complexity. This would be difficult to manage and could lead to a breakdown in narrative structure, leading to the loss of story identity (the hook/tag whatever you pitched the project as anyway).

Narrative Mapping 

There are many different structures that are used in game narrative. Below I have used models to illustrate this quickly.All Diagrams used for reference only and are attributed to Kano Ashwell, S. 2015.

The Time Cave

Copyright Kano Aswell, S


Copyright Kano Aswell, S

Branch and Bottleneck

Copyright Kano Aswell, S


Copyright Kano Aswell, S

Open Map

Copyright Kano Aswell, S

Sorting Hat (Branch & Bottleneck/gauntlet)

Floating Modules (non-structure)

Loop and Grow 

Copyright Kano Aswell, S  

I never lift content from another blog but I needed some visual reference on here. Kano Ashwell has a much better grasp of game narrative structure but as I would never use any content on my blog without fair attribution (as I’d hate that to happen to my work) please check out These Heterogeneous Tasks Blog for a much more complete description.


In Lunarium, I have been working with a model that includes a linear structure to begin with in the labyrinth, which then becomes an open world, then branches into two paths which then can be open world but ultimately ends in two endings.

©️ 2020, Sarra Hornby. All rights reserved.
Sarra Hornby 
Experimental VR Exhibition of Works

©️ 2020, Sarra Hornby. All rights reserved.
Loop and Grow Narrative Model 

AI Narrative Modelling

There are many really interesting things about the Charisma.AI app, the primary one is that the app makes it easier to write intuitive interactive stories. However, the app also helps to analyse the effectiveness of the narrative structure. This can be used to great effect when identifying which parts work well and which need improvement. It could also be used to modify a story to amplify an experience. This analysis is critical if it is essential that key plot points are experienced by participant/player/reader/audience.

Next Steps 

After much research in which I have attempted to better understand narrative, VR Narrative and AI narrative; I’m going to begin conducting sone practical experiments.

I’m going to begin by properly reading the Charisma docs and then apply this to a really simple (and I mean simple) narrative, within the Lunarium world.

So for example, I could begin with the Mystic of Lunarium’s backstory.

Why did she imprison innocent young teenagers in her sinister and aboding world?

I could then think about the Experimental VR Narrative and consider how AI might assist in this narrative experience … (This is a lot more tricky).


Recommended Reading:
Story by McKee
Save the Cat by Blake Snyder

Evaluation of an automatically-constructed graph-based representation for interactive narrative
Accessed 28/05/2020

Kano Ashwell, S. 2015. Standard Patterns in Choice-Based Games.

  2. Accessed 28/05/2020 

AI Narrative: A History (BrainStorm Examples)

When applying AI narrative in a project, it seems important to first look at the narrative of AI narrative. That is to say, the historical context through which AI Narrative is viewed as this could inform how it is used in practical terms.

What is AI? Where did it originate? Has the description evolved ? How is AI used to form narrative in media ? How is it used to form the media through narrative ?

I’m going to look at this pretty broadly and identify some examples that seem important to me (which are by no means exhaustive.. and may even seem off topic).



In Absentia ©️ Brothers Quay
Automatic Writing ©️William Kentridge 

Wall-e ©️ Disney 


Short Circuit ©️ TriStar Pictures (1986) 

Uncanny Valley 

Polar Express ©️ Warner Bros Pictures 


A Cyborg Manifesto ©️ Donna Harraway 
Full Text Available Here—-_a_cyborg_manifesto_science_technology_and_socialist-feminism_in_the_….pdf

And the following article in Medium.

The Turin Test

Cybernetic (Systems Control)

Cybernetics or Communication and Control in Animals & The Machine ©️ Norbert Weiner 

Full copy of the original Text is available here

2nd Order Cybernetics  

Quantum Mechanics : Heisenberg’s Uncertainty Principle, two slits experiments and the Wave/Particle Collapse. 

©️The information Philosopher 

The Double Bind ©️ Gregory Bateson 
(Narrative Designers should never place a player in a double bind!)


The Sage Handbook of Global Policing

Covid – 19 Societal Control – Policing by Consent. 


In the same way that typing google into a google breaks the internet (Joke) typing Netflix AI into google to search all films representing AI on Netflix leads to Netflix use of AI (well it amused me..)

I recently watched 2036 Origin Unknown which took a poignant look at humanity and AI.

And it reminded me of HAL 9000.

2001: A Space Odyssey ©️ Stanley Kubrick 


IBM Deep Blue Chess Game (1997) 
©️ Stan Honda/AFP/Getty Images
(For Academic Purposes Only) 

Alpha Zero ©️ Google Deep Mind 
A.I assisted narrative 

The Standard News Article

Some of these examples are pretty negative and it’s not my intention to cast a negative light on AI, I’m just trying to understand the narrative that surrounds AI and try to understand it better. 

References (TBC)

AI Narrative : Paper

Strange as it may seem, I’m going to begin my track through AI narrative history with the definition of animation as ‘the endowing of life and the endowing of motion’ (Chodlodenko, 2007). I begin here because the evolving definition of Animation as a practice in which the animator is the author of the animated world and everything in it is relevant to AI narrative in which the author has less control over the narrative aspect of the animated world.


I’m also going to talk about cyborg, Donna Harraway.

cybernetics, systems design and how Crowds are controlled (Policing and the pandemic).

2nd order cybernetics, the observer observing the observed and the particle/wave collapse.

Film Studies
Chodlodenko, returns to this topic again, referring to the way ‘animation as giving life to’ is Animation as film and media studies ‘blind spot’. (Chodlodenko, A. 2015).

First is my perennial annoyance—when attending conferences, lectures, etc., across all disciplines, including Film Studies—and hearing the words ‘animation’, ‘animate’ and cognate terms and phrases, like some thing ‘has a life of its own’, used all the time, but without any awareness, much less any address and interrogation, of them in their own right. They’re just part of the woodwork. I wanted to bring this discursive, but not only discursive, ‘blind spot’ regarding animation to light. (Cholodenko 2007a, p. 15)[1]

Although, he goes on to argue for art as a form of animation, for me this discourse aligns with my own intense irritation that the definition of Animation at that time was defined by theoreticians and didn’t resonate with my own experience of it which was that it was an immersive process, in which I was nested in several worlds simultaneously. The physical everyday world, the mental model of that world, my imagined world, space in which I created my imagined world where I engaged with the process of explicating that which was implicit. I did not give life or motion to my creations. That was not what it was about. It was about sharing my worlds with others as one might with any art but using the tools and processes I had, I produced something that came close to these worlds.

Game Studies 
The limitation was that the audience was always on the outside. However, now due to the wonders of modern technology, the door was opened and viewers can now be participants, immersed in these worlds. So my personal definition is about the immersive-ness of Animation, not its ‘anima/animus’. I think that anywhere in which the makers of art are as loud, if not louder (due to the weight of theoretician theory history) rather than the the voice of the theoretician as consumers of art (even if their interview skills are amazing), is a good thing. Game theory may have this for now.

Intelligence, complex information and interaction with a seeming ‘life force’ that is autonomous.

The recursive nature of making stop motion animation can be quite obsessive although satisfying is completely immersive; but for me it is not a very balanced way of living and creating art. The attention to detail to a myopic extent can be entirely exhausting and i often found myself having to keep something back in reserve.

I see now that leaning on automation, in certain parts of the production is a healthier working practice than trying to control all aspects of the world creation to this sane myopic degree. Unity is one of these tools that will help in this automation and through which I can hvae a high degree of control.

In terms of narrative, I see AI as a tool of equal benefit because it will allow greater audience participation, allowing for uncertainty and chance to enter the narrative animation process. Releasing authorship and control is a good thing. It might not be what the great tones of history which point to individual authorship of genius but it can be a social creation, an intimate exchange and a much healthier state for the artist to occupy.

Turin Test 
Measure of successful A.I.


Cholodenko, A. (2007d). ‘Speculations on the Animatic Automaton’, The Illusion of Life 2: More Essays on Animation, Power Publications, Sydney. Pp 486-528
Accessed 26/05/2020

Chodlodenko, A (2015)
Accessed 26/05/2020

AI Narrative: Modelling ‘mind’

What is narrative and how is it structured?

When I think about this, I am reminded of the helix project. Part of this experimental work was focused on trying to find a way to capture, organise, connect and disseminate, practice-based research.

Helix App Concept

At that time, when trying to represent the structure and process of mind, I turned to psychology. I used mental modelling theory (Craik, k, Johnson-Laird, P, Taversky,B) to Provide terms to help articulate my concept and inform the visual representation of mind 
I noted that when talking to people about how they think, I asked them to recall memories and describe how they did this. Some described a pathway, a cycle, a spiral or a labyrinth. Others described an experience of running and catching ideas, some playing golf but these activities were usually constructed in a world model of some sort. 
For me, at that time, I aligned my model as a cloud from which my ‘timeline’ was drawn as a spiral model. Through which I could look down and pick out key moments back in time. My memories were as ‘bubbles’ within a cloud. I could connect these together to recall a memory and was through this process that a narrative could be formed as a linear, sequential story line. This seemed limited and I wanted the audience to be able to see my model and have free access to my memories (and research). 
(I will publish the animation for Helix shortly).

I found that without having an image or physical/digital representation of my mental model, I would gesture and act out the mental model by using my hands and body, my finger pointing to an invisible structure in the air. 
This made me think about how I could give users the option of choosing their own model through which they could embed their own memories. 
3 months ago, I thought about how to represent the development of my experimental work. I considered how I might create this as a physical and virtual exhibition. 
I represented my ‘mind’ as a spiral along which a visitor could walk. 

I placed some examples of my work on the path. The visitors would have had the option of walking up and down the path. 

This was concept was further developed as a double helix with joining tunnels. This meant that  visitors could change paths if they chose to. This form was structured into the Lunarium project to give players the option to change their minds and ultimately their final ending. 

Research indicates that there is a way to effectively direct the story through spatial design and now I am thinking about how I can control instances of character interaction. I am thinking about these key moments, their narrative weight and whether I need or want, full control.

Questions such as ‘What do we want players to experience? What story would we like them to hear? Will we need to control the storyline in some parts? How might this be controlled and could this be less invasive? How might we create the right conditions for a ‘naturalised’ Interaction? At what point might the narrative breakdown?, are significant.

When it comes to ‘controlling’ players experience, the notion of control undoubtably raises questions of ethics and best practice, but I think AI appeals to me because the players have more control over their experience. In the next phase of research, I am going to examine Narrative structure, VR Narrative in general and specifically AI Narrative design. To do this, I need to take a closer look at some contemporary examples.