AI Art Generators: A Rorschach Test For Writers



Original Source Here

AI Art Generators: A Rorschach Test For Writers

Text-to-image models can be powerful tools for any writer seeking to understand the behaviours, fears, and beliefs impacting their practice.

‘/imagine: multicoloured ink blot on a computer screen resembling two people looking at each other, a computer keyboard in the foreground’, courtesy of Adam Summers and Midjourney

In the last year, AI art generators (AAGs) have become more accessible. Projects such as DALL·E and Midjourney create novel images from text-based prompts provided by the user. Examples can be found throughout this article, captioned with their starting commands.

Although their relevance to writers might not be obvious, AAGs open up exciting possibilities. Interactions with these models can illuminate the emotional and intellectual forces shaping our work, allowing us to be more mindful, courageous, and prolific in our creative practices. Considering the dynamics of these exchanges can be a powerful catalyst for creative growth.

Here, we’ll explore a collaboration with Midjourney, examining how these models can empower us to build a more joyful and fulfilling writing practice.

The ghost in the machine

‘/imagine: human silhouette in computer code’, courtesy of Adam Summers and Midjourney

The introspective potential of AAGs lies in their capacity to showcase an iterative creative practice that feels recognisably ‘human’ but resists much of the control we might wish to exercise. Held at a distance, we engage in creative correspondence with the AI, sending them our inputs and receiving their iterative responses. From these outputs, we can infer elements of their interpretation and decision-making, but we are limited in our ability to shape the result.

While AAGs are informed by human artistic expression, they are not human. Understanding this does not negate our compulsion to relate to these models in human terms. The urge to wrap humanity around AAGs places us in dialogue with a spectral collaborator, familiar enough to invite comparison with ourselves and our audiences yet lacking the ego necessary for collaborative symmetry. It is this paradox, the simultaneous absence and presence of another party, the combination of intimacy and remoteness, that makes AAGs fascinating barometers for the needs, anxieties, and assumptions underpinning our creative approaches.

The possibility of being misunderstood

‘/imagine: a fork in the road’, courtesy of Adam Summers and Midjourney

By decoding words to generate images, AAGs illuminate a vital, though often invisible, component of the author-reader transaction. Sitting beyond the writer’s agency, the interpretation of our work can be a daunting consideration. Seeing what others do with our words can be thrilling and terrifying in equal measure. Our reactions to this handover reveal much about our comfort with being misunderstood.

Unlike a keyword-powered image search, AAG collaborations show our words being refracted through vast webs of associations and references. It is here that these models display patterns of association and decision-making that feel most human, with a fidelity and transparency sufficient to elicit a sense of intrusion. I was drawn to consider how my words function for other people and how representative the model’s activities are of that process. Immediately apparent were the limits of my authorial control: the images and ideas I package in my words may not be the ones unpacked by their recipients. While AAGs do not profess to emulate a human interpretation of language, they expose a ubiquitous lie of convenience, that our words communicate exactly what we want to say.

The fear that others will misunderstand or fail to believe us is almost universal, threatening our feelings of identity, safety, and interpersonal connection. When preparing an AAG prompt, we engage with the possibility of seeing our words decoded in unexpected ways. As writers, a degree of ambiguity is the price of admission for working with an associatively rich medium. Discomfort with this trade-off usually manifests in impulses to exert control, adding resistance and depleting valuable resources. If we fixate on being understood, we can over-privilege the perspectives of our audience and believe them to be more definitive than our artistic intentions. Viewing our works as prompts, like those provided to an AAG, can reframe reactions to our writing as responses, rather than definitive interpretations.

Unpicking self-conscious tendencies in a creative practice

‘/imagine: self-conscious artist with audience’, courtesy of Adam Summers and Midjourney

In their first pass of the prompt, AAGs generate multiple images for refinement. In Midjourney, this initial set contains four responses, each honed multiple times before further input is requested. The AI begins with an inchoate concept and explores it, finding aesthetic overlaps and sketching out details in unresolved areas. This isn’t just adding detail to a fixed interpretation; there’s evidence of bold, sometimes destructive, decision-making at each step. For a prompt containing the word ‘buried’, the AI repeatedly embellished the silhouette of a figure by a graveside, only to replace them with a tombstone. The AI is certainly not afraid to ‘kill its darlings’.

‘/imagine: excavation of a buried sadness’, courtesy of Adam Summers and Midjourney

Watching the AI work is a rare glimpse into the messy, failure-ridden majesty of an effective creative process. Here, there is no ego obscuring the details and we see an artistic exploration unhindered by self-doubt or fears of judgment. I often feel my creative process is a far cry from the tidy, optimised practices used by other writers. Watching the AI’s approach was reassuring in a way I haven’t felt before. It validated my idiosyncrasies so thoroughly that I was prompted to consider why I wasn’t being more adventurous. While various factors were at play, the primary anxiety limiting my self-expression was a fear of creating something deemed uninspired or inappropriate, even as an intermediary step.

Imagined audiences can occupy considerable space in our creative practice. Any sense of surveillance can pull our focus from artistic expression to satisfying the imagined tastes of others. This compromise limits our capacity for risk-taking and play. My creative projects have always felt monitored by the eyes of an imagined, unsympathetic audience. I fear that, at any moment, my work might be snatched away and used to pass judgment on my worth. As such, creative dead-ends, misfires, and incomplete works transform from essential by-products into sources of danger. Since my experiences with Midjourney, I’ve committed to evicting this imaginary audience from much of my writing practice. Breaking this lifelong habit will take time, but an awareness of my impulse to people-please has already made me more deliberate in prioritizing my interests.

The power of making the ‘wrong’ decision

‘/imagine: ambiguous’’, courtesy of Adam Summers and Midjourney

Throughout the model’s artistic iterations, I was surprised by the variability in my preferred result at each stage. The least promising of the initial blurs could resolve into something transcendent, while early favourites would often be derailed. The process illustrates that the value of creative decisions is significantly shaped by the choices made further downstream.

It’s easy to lose momentum by fixating on the ‘best’ path forward. We often imagine that we’ll make optimal choices going ahead, and that a poor selection now will waste time or limit the quality of the work. Harder to see is the cost of indecision and the creative possibilities of the ‘wrong’ choice. Prolonged indecision squanders valuable momentum, replacing our excitement and curiosity with frustration and a sense of impasse. The successes of Midjourney present a convincing argument that making a decision is often more important than the decision itself.

External validation and our need for control

‘/imagine: person connected to their artwork’, courtesy of Adam Summers and Midjourney

With Midjourney, it’s the user who decides when the piece is ‘finished’. Selecting our preferred result, we become the co-author of an artwork beyond the technical abilities of the average user. I was surprised by my reaction to these works. Objectively, they were excellent, remarkable in their aesthetic and symbolic richness. Subjectively, they made me uneasy. I didn’t know how to relate to them or how they were related to me. I was unable to experience a convincing sense of authorship, to connect the quality of these outputs to my abilities and sense of worth. My involvement felt too asymmetrical, too detached for any entitlement to attribution.

Artistic control serves to facilitate and increase our proximity to favourable outcomes, while mitigating and increasing our distance from unfavourable results. Here, proximity and distance encompass concepts of ownership, attribution, accountability, and personal association. As creators, we all have a different relationship between our degree of creative control and our sense of connection to resulting successes. Some can delegate almost every creative activity and decision, yet retain a strong sense of ownership over the final product. Others may find that relinquishing even a little control severs any connection they feel to the result. The tendency of these relationships to invert when considering accountability for negative outcomes reveals the importance of our self-worth in shaping these dynamics.

If we have low self-esteem and conflate the reception of our work with our worthiness, we may have an appetite for control that hinders our creative practices. Working alone, our frustrations with the limits of our control may be directed inward, framed as failings that further evidence our deficiencies. In collaborations, what should be a marriage of ideas and talents can become an acrimonious battleground. Control is essential to a productive artistic practice, but we must exercise it judiciously, and accept its limits. This will always be more challenging when our self-worth is at stake. My interactions with AAGs have revealed that I still have work to do in uncoupling my worth from my work.

Barriers to entry

‘/imagine: walled off’, courtesy of Adam Summers and Midjourney

A common reaction to AI collaboration is that it feels ‘too easy’. Perhaps more interesting than defining what we mean by ‘easy’ is exploring the underlying belief that creativity needs to be difficult. If this is true, how hard-fought do our victories need to be and why? Some of our respect for creative endeavours lies in their barriers to entry, viewing the challenges surmounted as useful impediments to the dissemination of underdeveloped ideas. But does creating need to be difficult?

I believe that art concerns the communication of ideas. Developing an innovative, well-considered concept can be the hardest part of any creative practice. If someone has a great idea but needs assistance to communicate it, why should they be denied the opportunity? I’m not suggesting we abandon or devalue the crafts of working with various media, but we should never look down on those requiring assistance to realise their vision. As someone with ADHD, recent advances in AI-powered transcription have removed significant barriers in my writing practice. What one person may regard as ‘too easy’ can be another person’s lifeline. If the result contains the concepts you wanted to communicate, then it’s yours. In a world where everyone is empowered to create, the content of our work becomes paramount, and there are gatekeepers for whom this is a scary prospect.

Practical applications for writing

‘/imagine: writing with AI’, courtesy of by Adam Summers and Midjourney

While much of this article has focused on the introspective power of AAGs, I’ll conclude with some thoughts about their practical applications for writers. Here, the symmetry of their activities with our own and their talents for association-forming and decision-making make them powerful tools for any visually-inclined author.

Escaping creative depletion often requires seeking out inspiration. Unfortunately, our reward circuitry may be deadened to the promise of new ideas, and we may struggle to make connections between them. In this context, AAGs can offer a novel take on an overly familiar concept. Using the AI to explore the space around your idea can be an effective, low-commitment exercise to recharge your creative batteries.

Creative projects are typically decision-intensive. While some of these choices are fundamental to our vision for the work, many are arbitrary and contribute to a framework for our main ideas. Unfortunately, even minor decisions deplete the same reserves of executive function needed for more meaningful choices. Delegating arbitrary decision-making to a system that doesn’t experience fatigue can conserve mental and emotional resources that are better deployed elsewhere. As writers, we need to transform AAG outputs into words if we wish to use them. The necessity of this creative intervention can strengthen our sense of ownership for these more collaborative elements. Spared from the depletion of uninvested decision-making, we can attend to meaningful choices, and make faster, more satisfying progress.

One final prompt

‘/imagine: Rorschach blot that resembles two people in silhouette facing each other’, courtesy of Adam Summers and Midjourney

AAGs have a surprising utility for writers, offering a wealth of opportunities for introspection, inspiration, and improved accessibility. These models showcase an iterative, recognisably human creative practice unshielded by ego and unencumbered by anxieties. Functioning like a Rorschach test, our impulse to compare what we see with our own experiences can illuminate the forces shaping our creative endeavours. Considering the impact of these dynamics on our artistic expression empowers us to be more deliberate in refining our writing practice.

AI/ML

Trending AI/ML Article Identified & Digested via Granola by Ramsey Elbasheer; a Machine-Driven RSS Bot

%d bloggers like this: