AI and Scribing: An Infinite Loop
By Kelvy Bird and Emma Dulski
Why inquire into the intersection of Artificial Intelligence (AI) and scribing? Because the relational loop between humans and technology is an interface that is here to stay, and our growth depends on comprehension and adaptability with this juncture.
AI is a current technology that: “leverages computers and machines to mimic the problem-solving and decision-making capabilities of the human mind… [AI algorithms seek] to create expert systems which make predictions or classifications based on input data.”¹ Scribing — visually representing ideas while people talk — is a practice that involves deep listening, sense-making, and drawing. It takes place within a social field — a felt space amongst humans that affects “the way we think, feel, and act.”²
Why now? Well, this is one of the hottest topics of the day. Everywhere we turn, regardless of country or sector or scale of conversation, it seems like people are trying to figure out what AI means to them, personally and professionally. It is timely and prudent to inquire into the technology and to understand how its amplification might have social and spiritual implications for generations to come.
Of course, many people are concerned about how AI will affect labor markets and will replace work that is currently dependent on human capacities. Robots moving products in an Amazon warehouse is one example, removing the human labor previously required in the distribution process. Self-driving cars are another example, replacing the need for taxi drivers and even ride-sharing services. Our practice and profession of scribing, also known as graphic recording or facilitation, is no exception. When will AI have the capacity to create a drawing that looks like the below image, based on a detailed request?
We guess that it will happen within ~6 months of today, certainly within a year. Will scribes become irrelevant? We think not. Yet, how we work with the knowledge of new technology will determine our orientation as we approach our craft, and thus how we evolve and thrive or become obsolete. How we approach our craft will influence the way we show up in rooms, attend within communal contexts, choose what to represent, and ultimately draw. How we draw will influence what a group sees and how they might, in turn, understand our emerging reality.
Some background… We (Kelvy and Emma) recently hosted two online calls to explore this topic in more depth, and the following thinking comes from those spaces.³ The calls were part of an 11-month series of conversations on “Scribing Applied To…” various subjects. To reach all time zones, the calls have been offered at 11 am and 8 pm Eastern Time, except for the third call, where special guest Danielle (Wella) Thurlow joined us to offer her unique perspective on the topic of decolonization. The other topics we discussed were climate change, education, and integration of the first three sessions. The structure of each 60-minute call has generally been: the framing of content, breakouts of ~3 people, then open dialogue. On average, ~30 people have joined each round (some live and some reviewing materials asynchronously), from ~4–12 countries around the world, ranging from the ages of ~20–70.
For this call on AI, we received several new registrations and inquiries, confirming that scribes not previously in the conversation series were curious to explore this intersection. To begin each call, we summarized four major themes that emerged rom the past four sessions. These topics helped us ground, and bring the collective knowledge forward, as we stepped into new territory. Our four main themes were:
- Sitting with discomfort. How can we grow with creative tension, the gap between our capacity and aspiration? What do we do with unfamiliar or troubling content?
- Orientation and posture. Where do we place ourselves physically and energetically in a room, with a group? How are we part of the change-making process?
- Safety. How do we create spaces with psychological safety, where open curiosity and vulnerability are welcome?
- Redefining our role. What would it be like to shift from pure witnesses/writers to active participants? How can we hold space for others to come in?
We slightly changed the format between the two calls. Since the second call typically has fewer participants due to the time zone, we seeded the container with inputs from the first, while still providing ample space for the group’s exploration. The following summary includes the voices from both calls, integrated thematically.
The Potential of AI for Scribing
Although the benefits of AI were not the first theme to arise from the conversation, several comments highlighted positive uses of this technology. AI can ease our work. Olivier mentioned, “AI can help alleviate gaps in capturing information,” specifically useful in the context of scribing. Dzvenyslava added, “It can do amazing analyses and filtering, including pointing out unexpected connections or choices that are hidden in ‘big data’.” These filing, filtering, and compiling capabilities expand the scale of what we can process as scribes. With the assistance of AI, there is potential to synthesize even more information from larger groups or across longer time spans. Michele expressed excitement “about feeding in observations to AI and receiving images” in initial experimentation and Kate added that using AI as a tool could “be freeing, like when digital scribing started, which opened the possibility to cross time zones and places.”
We also explored what it might mean to train AI to work alongside us as scribes, like Matisse orchestrating a room-sized collage, where “paper scraps were retrieved, pieced together and meticulously pinned in place according to the artist’s instructions.”⁴ Or “like the masters, who had a workshop of artists doing paintings, sculpture in their style,” suggested David. Returning to the four major themes from the conversation series, what can we teach AI about our “posture” as scribes and how can we train it to support us through our work? Would we listen as AI draws? Would we scribe while AI recognizes repeating themes in the conversation? If a scribe is a conductor, which instruments can AI learn to play, giving us more capacity to enrich the melody of the group?
Questions That Arise
Despite these benefits, the majority of our conversation time was spent talking about the current deficits of AI, especially related to inherently human capacities. AI cannot yet feel, forgive, or sense. Gabriel shared: “As facilitators we try to seek and foster spaces of silence and presence to listen not to what is said, but what is revealed… from the energy that you feel,” pointing to the uniquely human skills required to hold space for emergence. Jean-Baptiste asked: “How can AI interpret/feel the silence in the room? How can AI feel the spirit/emotion in the room?” What would it mean for AI to passively observe instead of interpreting from direct inputs? How will this technology evolve to mirror our listening and sensing skills? When will AI develop a sensitivity to the wide range of human emotions? Geisa and Kate agreed that “It just doesn’t have a heart.”
“AI is still based on 1’s and 0’s. As scribes, we listen to the tone and pitch, and energy; when they change and shift even so slightly from moment to moment, we can sense and notice the difference. AI is only learning through data,” Vivien stated. How humans attune allows us to notice micro shifts that may be lost in the methods AI uses to collect data. Gabriel extended the notion with: “AI is currently learning from what is above the surface of the iceberg,” meaning AI can only interpret the visible data, not the structures, thinking, and sources of inspiration behind the data.⁵ Yolanda added: “Sometimes as humans, we do not say what we think,” highlighting that scribes “can still observe the whole system, and AI cannot yet do that.” Scribes pull together spoken and unspoken signals to create a reflection of the whole. Will AI be able to use more than one sense? When we see someone shift uncomfortably in their seat, or when we notice a group relaxing over lunch, we interpret these behaviors through our own human experiences. We know what it is like to get distracted, to be unsettled, to eat a delicious meal. How will AI account for these embodied sensations?
Along with the positive human qualities, human error, and imperfection also rose to the surface. Olaf contributed: “What AI cannot [do] is ‘irrationally’ fail and get the concept of shame or ‘forgiveness.’” We mess up, work through challenges, reconcile, and that is what makes us more resilient as people. Without moments of conflict, we cannot figure out how our differences can help us come together. As humans, we can also more easily forgive one another with the knowledge that our mistakes arise from similar faults originating from our human nature. AI will never fail because of inherently human qualities, so what does it mean when AI makes a mistake? What does it look like to work through a process of forgiveness with technology?
One unsettling question about AI is the aspect of morality. Olivier asked, “What is its moral compass?” There are unknown elements in both human and machine learning; where they differ are their parameters and boundaries. Whereas humans have a moral center to anchor ourselves through changes, the parameters of AI cannot yet replicate the complicated web of culture, socialization, and experience that inform our moral compasses. This moral center informs how we learn and integrate new information, impacting the choices we make. Without this grounding, AI cannot fully participate in/contribute to our human social system because it cannot comprehend the unspoken influences that shape our world.
Without transparency into the inner workings of AI, it is also difficult to navigate the topic of authenticity. When AI generates a new piece of artwork or writes an essay or lyrics to a song, it pulls from thousands upon thousands of data points from the internet. Does this process constitute ownership of the work? Is AI now an artist? Or is AI-generated content just a conglomerated collage of all that currently exists? How is our art-making process as humans different? Do you need to see the hand of the artist for it to be considered art? As mentioned earlier, AI on a vast scale does what we scribes do on a smaller scale: listen to an array of information and synthesize it into a succinct output. The difference is, as scribes, we never pretend to own the information, we only craft the way it comes together. Can the same be said for AI?
Of particular note is the insight that as AI learns from online data, it will absorb both truth and falsehoods, light and shadow, well-being and wounding. Dzvenyslava observed that our collective awareness of AI coincides with a collective awakening around trauma. She asked, “Will we grow/evolve first? Or will AI first connect with our trauma?” The technology is “based on our human biases, and it still relies on human input” added Sofi. As general users (not the developers or early adopters) we probably don’t know the range of inputs AI is accessing, therefore the results that something like Chat GPT produces are not trustworthy and require perspective and judgment.⁶ What is the danger in trusting information based on imbalanced inputs of trauma?
Trust came up as a key element to cultivate while adapting to a world with AI — trust between humans and within society. Jayce offered: “We all need each other, to keep understanding one another.” Olaf added: “At a [factual level] I am not afraid of AI. My fear is for society as a whole,” voicing a concern others shared, that a line between truth and falsehood will only increase existing polarization and relational breakdown. He continued: “We need to trust humans more.” We might also consider slowly trusting the technology itself, maintaining an open attitude of cautious curiosity as it evolves as an inextricable part of our lives. However, we cannot rely on technology for reconciliation or collective healing. Our methods of untangling systemic oppression — like leveling power imbalances, for example — help us build trust in relation to one another.
While learning to enfold AI, we can still appreciate the mystery and imperfection of being human. Yolanda shared a story of a friend who scribed for an AI company where, during a panel, someone said that they “[cherished the imperfection of the scribing, as a quality of the human being.]” Yolanda went on to offer that Chinese calligraphy — with interconnected characters — becomes “dead” when transformed into a font. “In calligraphy, each character is not independent but rather part of an organic system” with the amount of space and thickness of the brush strokes adjusted to the qualities of each symbol. Sita added that“if you combine some of the algorithms, you might very well get some human-like hand in the images, or it could be trained to be imperfect—spelling things incorrectly to mimic human intervention.” Gabriel suggested that subtle qualities of variance convey an “essence of being” that scribing might preserve as digital landscapes expand.
Remaining Inquiries
What is our relationship with AI? The unspoken underpinning to our conversation was the question of how we approach our relationship with this emerging technology. Is it a confrontational and competitive one, or is it co-creative and collaborative? Maybe a bit of both, but it’s not yet clear what that might look like. There seems to be a lack of transparency to average users about how quickly AI is learning and growing. The data it collects is both intentional and unintentional; while a Chat GPT query provides answers for the user, it also gathers data about the user. For example, “Where can I find an outdoorsy Father’s Day gift?” indicates a consumer-related search, and identifies where in the world certain holidays are observed, when gifts are customary in which cultures, what types of gifts might be usual in those places, which device/browser a consumer asking this information would use, and much more. The point is that we do not entirely know what we are encountering; in a way, it’s alien. What is our approach to something that could drastically alter our lives?
How does this relationship evolve when humans largely live within boundaries of time and place, while AI is currently unbounded? AI can source data from anywhere online, from any hard-wired network and internet-connected machine. AI outputs information in every direction. Human scribes source data through a set of analog interdependencies. Even if we use a mobile phone to read an article online, we still hold a physical device. When drawing for a group of people — either in person or digitally — the data we receive and the information we generate are located in a 2-dimensional, visual plane. A drawing might convey a timeless feeling or energy, but at the end of the day, it’s a fixed artifact. AI has an ever-expanding spherical range, with fixed artifacts and unknown scope.
Can scribing play a role in the development of this relational intersection? Olivier spontaneously asked: “What’s the Chinese term for crisis and opportunity?” To which Jayce quickly responded: “There is a Chinese term for ‘Crisis is opportunity’ “危機就是轉機. Wei Ji = Crisis = 危機 Ji Yu = Opportunity =機遇. 危機就是轉機 Crisis is also an opportunity. Wei Ji = Crisis = 危機. zhuan ji= opportunity =轉機.”⁷ These are two sides of the same coin. As previously mentioned, how scribes draw — choosing what content to represent, highlighting patterns, emphasizing key points — influences the way the participant audience understands reality.⁸ If scribes consciously represent content anticipating that AI will access the drawing and learn from it, could they be part of some kind of awareness-based AI evolution?
Can scribing also encode and insulate shared human knowledge? Though this did not come up directly in either call, the line of thinking extends from our current inquiries. How can scribing help us to protect ourselves, given that our data is currently retrieved, stored, circulated, and leveraged without our consent? For example, designers have invented clothes that block technology from capturing facial recognition data. Kate Morales teaches in their programs how visual maps might not make sense to those in power but might hold sacred meaning for marginalized groups. Can scribing serve as a vehicle to ensure the safety of our communities from the malicious intentions of big tech?
What might an ecological approach look like? A subtle inquiry around AI’s comprehension came through Allie’s particularly deep, integrative lens. She reflected: “How we orient with it and attune to it informs how we relate to it and how it will learn from us — in a way like [how humans relate to and learn within] the social field, but now expanded to a socio-techno field.” She added: “We are human bodies and this is inviting us to be different; it’s like cultivating a new garden… How might an intentional, multi-dimensional, trans-contextual relationship with AI be evolved, so humans and AI can dance better together?” If we consider the spherical, almost octopus nature of AI, reaching and making connections in every direction, how can humans keep up? Can AI ever really be part of a social field, as we know it?
How can AI aggregate scribed information? Kate (Baxter) brought up the artist Refik Anadol, who “used artificial intelligence to reinterpret the metadata that comprises the online database of MoMA’s immense collection… The results are cool patterns that recall digitized versions of van Gogh paintings morphing into paintings that look like Monet’s, which then turn into de Koonings or Frankenthalers or Rothkos, into sandstorms and mashed potatoes and other things that sort of look like art.”⁹ Imagine what is possible through the metadata of thousands of digital records of scribed conversations from all over the world, from every sector, in dozens of languages, added online by a few thousand graphic recorders/scribes over the past 50 years, since the modern-day practice began? The data would be incredible, and the patterns…! It’s hard to fathom what might be revealed and what could be learned. Only a set of sophisticated algorithms could take on that kind of processing.
Lastly, what determines the quality of a scribed image generated by AI? Marie-Pascale posed: “Assume AI created an image that was ‘complete nonsense’ as it does sometimes. AI often ‘pretends’ to know; how do we know whether there is actual reasoning behind an abstract visual? Would we be able to distinguish ‘genuine scribing’ from ‘nonsense’ scribing? People would likely read meaning into either. Is it ‘bad,’ because it looks different then what humans produce? Is it ‘good’ when it looks similar to what we do? And if it sparks resonance with humans, is it even relevant?”
In conclusion, we wonder how can scribes reinforce the uniqueness of human experience in their images? What will be required on the part of each scribe to increasingly attune to spirit, to heart? And if our awareness can match this call, what will then shift in our tangible approach to drawing? Will we write fewer words and focus more on letting our imperfection shine? Perhaps that is a larger summons as AI weaves ever more deeply into our lives: be flawed, stay spacious, and celebrate the subtlety in each wavering mark.
Participants in the Conversations
Allie Middleton — USA
Crystal Huang -Taiwan
David Bennett — New Zealand
Dsvenyslava Novakivska — Ukraine
Emma Dulski — USA
Gabriel Tuan — USA
Geisa Paganini — Brazil
Jayce Pei Yu Lee — Taiwan
Jean-Baptiste Bonvalot — France
Kate Baxter — Australia
Kelvy Bird — USA
Michele Gauler — Germany
Mish Kinsella — USA
Olaf Baldini — Germany
Oliver Pesret — France
Sofi Donner — Uruguay
Vivien Leung — Germany
Yolanda Xu — China
With thanks for the additional thoughts that have been woven in from Lark Collective members: Kate Morales, Marie-Pascale Gafinen, and Sita Magnuson.
— — — — — — — — — — — — — — — — — — — — — — — — — — — -
[1] IBM. (n.d.). What is Artificial Intelligence (AI) ? IBM. https://www.ibm.com/topics/artificial-intelligence
[2] Pomeroy, E. (2022, May 18). Social Fields: Shifting the quality of our collective being. Field of the Future Blog. https://medium.com/presencing-institute-blog/social-fields-shifting-the-quality-of-our-collective-being-6bfc8de3afac
[3] See the end of the article for a complete list of participants in the June conversation.
[4] Spurling, H. (2014, March 29). Henri Matisse: Drawing with scissors. The Guardian. https://www.theguardian.com/artanddesign/2014/mar/29/henri-matisse-cutouts-tate-modern-drawing-scissors
[5] Scharmer, O. (2018). The Essentials of Theory U: Core Principles and Applications. Berrett-Koehler Publisher.
[6] “Chat GPT stands for Chat Generative Pre-Trained Transformer and was developed by an AI research company, Open AI. It is an artificial intelligence (AI) chatbot technology that can process our natural human language and generate a response. Simply put — you can ask Chat GPT a question, and it will give you an answer.” Drue, J. (2023, March 31). What is chat GPT? Lighthouse Guild. https://lighthouseguild.org/what-is-chat-gpt/
[7] “轉機 Zhuan Ji means turning point when you read it stands alone, but when you put these two characters(轉機 Zhuan Ji) with WeiJi (crisis) then there is this Chinese slang/term ‘危機就是轉機 Crisis is opportunity. 轉機 Zhuan Ji also means transfer (like how you need to transfer to another flight),” from Jayce.
[8] The use of “participant audience” marks a shifting frame of reference from “viewers” of art to active players in an artwork’s creation, to coincide with the shift from movements of “fine art” to “social art”. This is of note when considering the extension of this participatory approach with AI.
Bird, K. (2018). In Generative Scribing: A Social Art of the 21st Century (p. 6). PI Press. https://kelvybird.com/wp-content/uploads/2018/01/Bird_GenerativeScribing_Intro.pdf
[9] Saltz, J. (2023, February 22). Moma’s glorified lava lamp. Vulture. https://www.vulture.com/article/jerry-saltz-moma-refik-anadol-unsupervised.html
