The creative process & tools of the trade. Can GenAI become creative?
A stab at capturing the creative process.
When artists create, they let their inspiration and creativity guide their process, regardless of medium: A canvas and brushes are used to produce a painting, a camera is used to capture and freeze a significant moment, musical instruments are used to compose music, the human body is shaped and moved in ways to perform dance, a pen and paper are used to write and form poetry. Art is intrinsically a form of expression, sometimes with the focus on generating income, create beauty, to express emotions, to ?inspire or even to tell a story.
Because the process of being inspired or being creative cannot be captured in one single graph or process, because it means so many different things to different people and art and creation are ?incredibly versatile (our human brain works in billions of different “gray-matter-neuron-driven-ways”), I’ve only attempted to capture a high-level view of the creative process in the image below:
The “creation” component of this process is typically a human managed process, inspiration is not. Inspiration can’t be forced; you can try to be inspired by allowing yourself to immerse yourself in certain situations. For example, I typically get inspired from internal sources (the brain) while I’m exercising. If you ask me why that is, I can’t answer that question in a meaningful way, but it works – most of the time. I can also get inspired by other pieces of art, a composition, a photograph, nature itself – external sources. ?The human-led creative process is more “manageable”, the output is not.
Making the creative process more tangible.
Painting, drawing and wood carvings
I remember my father, who was an artist by trade, simply shred or throw away his art, because the output didn’t satisfy him. His criteria for satisfaction were unbeknownst to me. However, every night after dinner he escaped into his studio and spent hours and hours creating. Sometimes after weeks of shredding and re-building he would output something he deemed worthy to keep. This was his creative process, and it was always iterative and outcome focused.
Photographic creativity
Let’s align another real-life event with the creative process, In May 1957, Martin Luther King spoke – relating to the third anniversary of a legal case “Brown v. Board of Education” (1). The famous photographer Henri Cartier Bresson was present at the time, and captured the event. While taking pictures of Martin Luther King, Cartier Bresson was in turn photographed from the front by Bob Henriques while Martin Luther King can be seen from the back.
In line with the high-level creative diagram above, the inspiration for the photographers was the event itself (an external event), the creation was the initial output of the photos (the creative process). In the day of film and darkroom before the final output was presented, analysis and darkroom techniques as well as printing styles were all part of this creative process. (The various forms of output before the “final” result).
Music and creativity
Taking a peek at the incredibly creative mind of Trent Reznor, known for being the founder and creative brains of Nine Inch Nails. Reznor has now also moved into creating soundtracks for the film industry and won an Oscar for The Social Network (3) together with his long-time other half brain, Atticus Ross.
When Trent Reznor was recording an album through his then record label Interscope Records, Year Zero (4), he was touring most of the time. Not having access to his studio, he created most of the music using a laptop, leveraging an almost completely digital process. In an interview with “Kerrang” he stated: “I didn't have guitars around because it was too much hassle?... It was another creative limitation?... If I were in my studio, I would have done things the way I normally do them. But not having the ability to do that forced me into trying some things that were fun to do.” The inspiration for the creativity almost seems to have originated from a limitation on the tour bus. He had no other choice but to remain creative – external sources and events led to a nonstandard creative process – as well as incredible output.
During concerts, to promote the upcoming work, he would leave USB sticks with material from the unreleased Year Zero in certain venues. One such USB stick was found in a bathroom stall in Lisbon, another was found in Barcelona. Obviously, this material very rapidly found its way to the internet where it was virally shared. It led to a “cease and desist” letter from the Recording Association of America, even though Interscope records had already approved the promotional campaign. Trent Reznor in an interview with the Guardian: “The USB drive was simply a mechanism of leaking the music and data we wanted out there. The medium of the CD is outdated and irrelevant. It's really painfully obvious what people want?– DRM-free music they can do what they want with. If the greedy record industry would embrace that concept, I truly think people would pay for music and consume more of it.” Overall, an incredible new way to reach fans and get people excited about the upcoming album – a form of Guerilla Marketing. (5) Not only did Reznor release a superb album, he also transformed his creative process and shook the recording industry by introducing a complete shift into how music can be sourced and distributed. ?
The Year Zero album, the output of the creative process on the tour bus, led to the digitally remixed album “Year Zero Remixed” (6). Other musicians took Reznor’s raw digital work and gave it a spin of their own – originating into even better versions (updated output) of the original version (initial output). The album even contains the raw material of all songs for people at home to create yet another form of output. Creativity as a viral construct. (7)
领英推荐
How does GenAI fit in?
Let’s have a look at where we are with GenAI within this creative “artistic” process (I deliberately do NOT talk about coding / development processes ) I'm focusing on the creative aspect of GenAI.
GenAI is obviously incomparable to a human and can’t be inspired by itself, either through internal processes and events or external ones for that matter. Instead, consider the human the inspiration for GenAI – the human delivers the input and GenAI starts its creative process – leading to a multitude of types of output. Today, GenAI has already shown incredible creativity: creating poetry, writing theses, blending its massive source of imagery for marketing content, creating product photography for websites, generating lyrics for new songs – the possibilities are seemingly endless. But all of this creation is impossible without GenAI having access to data originated by humans. GenAI isn’t a creator in and by itself, it’s a blender and generator of human provided data stimulated by human created prompts.
Another area where GenAI is strong is in iterative improvements; it requires the human to step in and guide its creation process to take it further and continuously improve on its own output. It’s fast, creative in sourcing from its massive repositories and is a perfect aide to the human counterpart.
Using “traditional” AI to inform GenAI for the results of generated content would be a good example of an output-to-output enhancer. Let’s look at an example of a champion / challenger program. A marketing campaign is running in a sample sized test program. It has several pieces of content: copy and images. Based on acceptance rates, a winner can be declared based on this typical continuous test as well as values / success criteria set by the marketer.
The AI generates the metrics of the campaign and based on “success criteria thresholds” set by the marketer, over time, the champion is selected and can be presented to a larger audience. But wait – let’s not discard the underperforming content – why did it perform worse than the winner? Let GenAI decide upfront and during campaign the difference and variances of the content, compare it to the audience that did respond positively to the content and generate similar new content based on that winner. If content underperforms, it doesn’t have to be thrown out – it’s just as important to know why something didn’t work for a particular audience as to just accept that something else did. To improve GenAI’s copy and content it should learn from winners AND losers AND generate its content iteratively ongoingly – verified by the marketer and the data scientist.
This leads me to the second example – let GenAI “Run Rampant”. In the human creative process model, the human oversees the whole process (with the Inspiration component being unmanageable / uncontrollable). However, in the GenAI case, it’s the human and the massive data pool that can inspire the engine and let it create for the sake of creation. With the right initial prompting, let GenAI continue to provide variations and keep testing those in smaller randomly sampled audiences. While this is very similar to the first example, we don’t stop GenAI when the marketer is satisfied (because that’s still biased) but let it continue to bloom –verify the outputs and let it explain itself. Let GenAI be an “uncontrolled controlled” lab rat. I’d love for GenAI to create for the sake of creating – like an artist, for it to express itself, to inspire and to tell stories.
?
PS those lab-stories can be intriguing and frightening at the same time. See (8) – a good background overview on Loab, the internet sensation of late 2022. Wouldn’t it be intriguing to talk to GenAI and ask it about its reasoning of expressing itself the way it does? Artistically?
Sources in this article (all art owned / protected by the respective copyright owners):
(1) https://www.archives.gov/milestone-documents/brown-v-board-of-education: In this milestone decision,?the Supreme Court ruled that separating children in public schools on the basis of race was unconstitutional. It signaled the end of legalized racial segregation in the schools of the United States, overruling the "separate but equal" principle set forth in the 1896 Plessy v. Ferguson case. (2) https://www.magnumphotos.com/shop/collections/darkroom-prints/darkroom-prints-washington-d-c-1957/
Head of Sales @Pega | Executive MBA @MIT Sloan | Passionate about Empowering Organizations with Real-time Decisioning and Analytics
1 年You got me at Trent Reznor ??