Reinventing Creativity in the Age of Intelligent Machines
In this article we will briefly delve into my experience with generative AI in art and creativity, from early experiments with DeepDream and ArtBreeder, going through my early tests with OpenAI models. We will then reflect on the transformative impact of AI on artistic expression, without leaving behind the concerns of traditional artists.
Synthetic Visions
My adventure into the realm of generative AI began a few years ago when I started playing around with Google’s DeepDream, a tool that transformed your pictures into psychedelic and surreal digital artworks.
This early exposure was a window into how AI wasn’t just about data and prediction algorithms, but also had the potential to autonomously create ‘stuff’.
Back then it was nothing short of magic to my eyes.
My intrigue grew as I started playing around with Artbreeder, a platform that allows to blend images by leveraging Nvidia’s StyleGAN generative model.
The ease with which Artbreeder enabled the creation of complex visuals, also allowing you to ‘breed’ multiple images generated either by yourself or by other fellow art breeders was extremely fascinating.
My experience with ArtBreeder consolidated my foundational understanding of AI’s potential in visual art. I quickly realized that it wasn’t just a tool to play around with randomized image replication, but rather something that lowered the barriers of creative expression, also enabling a new form of collective art, capable of adding its own unpredictable twists to creative outputs of one or more people.
Experimenting with Language Models
As generative AI progressed, I witnessed and experimented with its various incarnations, each offering new perspectives and possibilities.
When I got access to the Beta of OpenAI’s GPT-2, and later GPT-3, my experiences went beyond visual art, entering the domain of language.
The models, particularly OpenAI’s DaVinci, had the potential to open up new avenues in human-machine collaboration, generating text that was often indistinguishable from human writing.
I remember creating my first rudimental chatbot in Python which I called ‘IOTA’ (Interactive Oracle for Textual Assistance) and starting to experiment with OpenAI models DaVinci and Curie. At that time, this allowed me to experience how such models, by ‘conjuring’ words, could answer complex questions, weave narratives, argue points, and even mimic literary styles.
Sadly, back then the API was quite expensive (especially when using DaVinci), so I couldn’t use it extensively, however, I still had the chance to play around for a bit and was totally blown away by the potential of such systems.
Addressing the Critical Voices
Despite I’ve been personally excited about generative AI for years, its commercialization starting from 2022 and its consequent introduction into the mainstream creative realm has not been without its controversies so far.
Many artists and writers express valid concerns, feeling that their very essence and livelihood are being challenged by this new wave of transformative technology.
Central to their unease is the feeling of being “ripped off” and, in a way, replaced by the machines. The models that power generative AI, including Large Language Models and various Generative Adversarial Networks for image-generation, are often trained on vast datasets that include works by human creators — often without explicit consent or compensation.
This raises a critical question: when an AI produces a piece of art or writing, whose creativity and ingenuity are we actually witnessing?
Additionally, the notion that creations by AI are not “real” art sparks a rather tricky debate.
Traditionalists argue that art, at its core, is an inherently human endeavor — a product of individual emotion, experience, and interpretation.
This sentiment taps into the deep-rooted fear that AI could dilute the uniqueness and value of human expression, transforming art into a deterministic product of algorithms, rather than an exploratory journey of human emotions and experiences.
There’s the concern that the reliance on AI for artistic creation could lead to a diminishment of skills and traditional techniques that artists have honed individually over years, and collectively over centuries. Skills in painting, sculpting, or writing are not just techniques but also encapsulate historical contexts, cultural narratives, and deep personal stories.
If such depth is sidelined by the ease and efficiency of AI, we might lose not just the techniques themselves but also the stories and heritage they carry.
This fear probably stems from the belief that art should be a manifestation of the artist’s inner world, a unique and intimate translation of their thoughts, feelings, perspectives and technique. When machines generate art, personal nuances and idiosyncrasies might be overshadowed or even replaced by the predictability and efficiency of machine learning models.
In addressing these fears, it’s crucial to strike a balance. While embracing the new horizons opened by AI, we must also preserve the irreplaceable elements that make human art so unique and valuable.
The ability to capture the depth, complexity, and unpredictability of human experience, ultimately can still remain firmly in the hands of human creators.
However, there will still be people with different levels of acceptance when it comes to new technologies, and this has to be acknowledged and respected, not suppressed.
To AI developers, I ask to seek for authors’ permission when extracting data and content for training purposes.
It is not fair that people’s hard work is utilized without their consent, in the same way it wouldn’t be fair if your code would be exploited and commercialized with a different label without your permission. I’m sure there will be creators who will be happy to donate their work so that it could become creative DNA to be utilized by millions of synthetic creators.
I believe that we can find harmony between data ethics and producing AI tools that are not only excellent but can also constantly enhance and challenge our creativity.
My current sentiment on Generative AI
Despite there being a lot of work to do on the front of legal frameworks and data ethics, my current view of generative AI is optimistic. I believe that when it comes to human creativity, AI will stand not as a usurper, but as a groundbreaking tool that will create a new artistic paradigm. Synthetic art will not replace, but rather exist in parallel with traditional art and even digital art; in the same way as photography has been doing for almost two hundreds years.
The dawn of photography in the 19th century did not replace traditional art such as painting or sculpting, but it provided a new tool to capture reality on a different medium and with different techniques.
In a similar way, I see generative AI as a tool that allows us to capture into the digital medium what is in our minds, either as writing, imagery, sounds or videos. And it does so, without the need for us to spend years mastering creative skills.
I see it lowering barriers to creative expression, democratizing art creation, and opening new avenues for exploration and innovation.
There will undoubtedly be casualties among digital artists and writers, particularly those who find it difficult to adapt to this new paradigm.
Traditional skills, while still valuable, may need to be augmented with an understanding and mastery of AI tools to remain relevant and competitive, expecially in the digital market. Creatives with a small audience who rely solely upon old methodologies may find themselves increasingly sidelined, as the creative world continues to evolve at a pace driven by technological advancement.
But this technology, in my eyes, is a liberator, not a conqueror of the artistic realm.
People who may not have had the means or the training to express themselves artistically now find in AI a partner that compensates for skill gaps and complements their ideas.
Abandon all hope, ye who enter here.” — Canto III, Divine Comedy by Dante Alighieri- Artwork generated with Midjourney, October 2023
Ultimately it will be all about niches of creators and consumers. Some of them will prefer painting, others Photoshop, and other ones Midjourney.
Long-Term Benefits of Generative AI
I think that we shouldn’t be focusing solely on the automation aspect of the technology, but start to collectively reflect on the long-term augmentation capabilities that will enable individuals to translate abstract concepts into ‘tangible’ creations, irrespective of their technical expertise.
The broad, long-term picture reveals a transformative shift in how we conceive and actualize creativity, inviting everyone to take part in a new era of artistic expression.
Take, for instance, the case of Large Language Models and their profound impact on linguistic barriers and learning challenges.
Dyslexic individuals, who might struggle with traditional writing and reading, are finding in ChatGPT a supportive tool that aids in organizing their thoughts, correcting spelling and grammatical errors, and improving overall readability of their text.
This assistance doesn’t just help correcting mistakes but it also gives them the confidence to express their ideas more fluently and effectively.
Previously, someone struggling with dyslexia might have never imagined becoming a professional writer or blogger. Now, thanks to generative AI, that’s a real possibility.
Similarly, non-native English speakers are harnessing the power of ChatGPT to bridge the language divide, enhancing their English proficiency and communication skills.
This opens up opportunities for them to share their thoughts, stories, and cultural perspectives more clearly with a global audience, something that was considerably more challenging before.
Their voices, previously marginalized due to language barriers, can now be heard and appreciated in broader circles.
In short, I think generative AI empowers the collective.
It’s about connecting more intellects and cultures, fostering a whole new level of collaboration that was once hindered by geographical and time constraints.
These effects could be long lasting and potentially life-changing for entire generations of individuals.
Closing Thoughts
My personal journey with generative AI has been full of surprises and revelations. Each new tool and technology opened doors to previously unimagined realms of creativity, and it’s continuously reshaping my understanding of what it means to create and be a creator in the digital age.
While some view these advances in AI with skepticism, fearing a diminution of traditional artistry and ethical concerns, I recognize that there is a spectrum of technology acceptance and also the importance of data ethics. I personally see generative AI as a broadening of the creative palette, an enabler that breaks down barriers and enhances human creativity.
As someone who has spent hours sketching and creating at an amateurish level, experimenting with platforms like ArtBreeder first and Midjourney and Dall-E then has been life-changing.
Imagine then, what could be achieved by those who have dedicated their entire lives to artistic pursuit, should they choose to leverage such tools to assist or enhance their artistic skills.
I believe that the involvement of machines in this new form of synthetic art will not be black and white, in the sense that there will be a machine involved or not in the artistic endeavor. The human experience is not made of 1s and 0s but it’s more like a spectrum; it’s analogic we could argue.
Synesthetic art and creativity, exactly because they are driven by humans, will reflect this gradient of possibilities.
The involvement of the machine will differ depending on the author, what they want to express and the way they intentionally chose to use AI tools for forging their creations.
They may be images and text completely generated by AI, while in others AI will be used only as a starting point, and assistant or an enhancer.
In some instances, AI might provide the initial brushstroke or concept, which is then refined and reimagined through human intuition and emotion.
In other cases, AI could be a collaborative partner, working alongside the creator to meld ideas and inspirations into something neither could achieve alone.
This gradient in the involvement of AI reflects the diverse spectrum of human creativity itself.
Just as no two artists paint the same way, no two creators will utilize AI in identical manners. Some may lean more heavily on these tools, using them to overcome technical limitations or to explore realms beyond their conventional skill set; others might use AI sparingly, perhaps for inspiration or to challenge their traditional methodologies; and someone else might blend LLMs and GAN models to craft complex, imaginative fictional worlds.
The beauty of this evolving form of art lies in its unpredictability and the infinite possibilities it opens up, while making creative expression accessible to everyone.
The dialogue between creator and creation, the exchange of ideas between the organic and the synthetic allows for an exploration that goes beyond the conventional boundaries of medium, style, or process.
Some see generative AI as laziness; to me it’s digital alchemy.
We will be able create colorful universes made of poetry, music and imagery intertwined all together.
And we will be able to do it on our own or together.
To artists, writers and creators everywhere, I extend an invitation: delve into the realm of generative AI.
Experiment, play around with this technology, uncover new techniques and use cases. Remain as curious and creative as you’ve always been.
This isn’t just a new tool; it’s a new paradigm.
Embrace it with the spirit of exploration and creativity that defines your art and your skill.
By working together with our new intelligent tools, we will give birth to creative hives of human beings and AI systems who co-create and tell stories about ourselves, our culture, our machines, the world we live in and ultimately our dreams.
If you enjoyed the read, I suggest reading my other piece about the dawn of symbiotic creation, published here on Brass for Brain.
The article contains a synthetic writing piece that I created some time ago by leveraging GPT-3. If curious to read more synthetic writings and poems with different degrees of AI involvement, I’ll be more than happy to publish more of them on Medium.
Cheers!
Thank you for lending your attention to me.
I hope this article contributed in some way to enrich your life, expand your knowledge and provide some value amidst the ocean of information we navigate daily.
Feel free to discuss my content with your friends and colleagues and also consider taking a brief break from screens after reading it.
Take a few deep breaths, stretch, and go for a short walk. Returning with a refreshed presence may help metabolize and integrate new insights.
Daniele
