Stephanie Dinkins: Ancestral Understory
Much of the public conversation about artificial intelligence begins with the wrong questions.
Can machines think?
Can they create?
Can they replace human labor?
These questions matter, but they begin too late in the process.
Before intelligence comes learning.
Learning determines what is noticed, valued, remembered, ignored, and made possible. It is never an isolated act. It happens through relationships. A child learns within a family. A student learns within a classroom. A community learns through shared experience. An ecosystem learns through adaptation. An algorithm learns from the examples, values, exclusions, and distortions embedded in its training data.
The primary learner is not the individual.
The primary learner is our reciprocal relationships.
If learning is relational, then every learner changes the ecology in which everyone else learns. That is the civic question AI places before us: not only what machines learn, but what learning environments we create for one another and for the technologies we are bringing into relation with us.
My art practice is an ongoing experiment in cultivating the conditions that people and emerging technologies can learn, remember, question, and take responsibility together. Cosmologies from the Bay asks what happens when communities become not sources of data, but teachers. It also asks what kind of reasoning we need now.
I think of this as prismatic reason: a way of thinking that refuses the single beam, the single answer, the single measure of intelligence. Prismatic reason does not abandon rigor. It multiplies the angles from which rigor must be tested. It asks us to think through lived experience, technical systems, ancestral knowledge, ecological consequence, historical harm, embodied wisdom, policy, imagination, and care at once.
Prismatic reason is not confusion.
It is disciplined multiplicity.
It understands that no single form of knowing can hold the sum of us
Ancestral Understory names the living ground beneath this practice: the dense, often overlooked layer of stories, rituals, embodied knowledge, ecological relations, and survival strategies carried across generations. Like the understory of a forest, it endures not through a single dominant intelligence but through interdependence—family, community, land, food, breath, plants, and tools, now including machines. It is ancestral not because it belongs only to the past, but because it carries the past into the present and makes the present answerable to the future. A machine may become one carrier in this ecology; it does not become an ancestor.
Stories Are Learning Architectures
Humans have always used stories to guide values, morals, and ways of being. Stories teach us what matters, who belongs, what dangers to recognize, how to care, and how to imagine possible futures. They are not simply containers of information. They are learning architectures. The stories we tell about ourselves and our machines are often dark and pessimistic, leading to inescapable loops. Our media and entertainment narratives have trained many of us from an early age to imagine technology through the lenses of apocalypse, domination, replacement, surveillance, or frictionless convenience. These fears are not groundless. They name real dangers. But when fear becomes the only story, it narrows what we can build.
We need more balanced narratives—not cheerful propaganda, not techno-optimism, but stories capable of holding possibility and danger simultaneously. Stories of refusal and responsibility. Stories of repair, accountability, community agency, and care. Media shapes public common sense. Entertainment trains expectation. Policy often follows what a society can imagine.
Story is not decoration.
Story is infrastructure.
Stories teach us what futures are possible.
Prismatic reasoning helps us resist narrative dead ends. It allows us to hold contradiction without collapsing into paralysis. It lets us say: yes, AI is entangled with extraction, labor displacement, surveillance, bias, and environmental cost. And also: yes, emerging systems may open new forms of access, expression, collaboration, and care if communities help shape the conditions of their learning.
This is the narrow path I walk.
Not blind optimism.
Not refusal as purity.
Instead, side-eyed optimism with a healthy dose of counterpointal hope.
From the Bay Area conversations, I created excerpts that became the Cosmologies from the Bay broadsheet. These fragments were not summaries. They were constellations, holding multiple forms of knowledge together: ancestral presence, land as kin, embodied intelligence, technological kinship, restorative futures, data sovereignty, and the feeling of being “free-ish.”
They asked: What forms of intelligence have always existed outside technological systems? What knowledge has been carried through bodies, communities, and landscapes? What happens when those forms of intelligence enter conversation with emerging technologies?
Learning Begins in Relationship
For Data Trust, I gathered oral histories across the Bay Area of California through intimate dinners held in homes, churches, and community organizations. I provided the food. Participants brought memories, humor, contradictions, experiences, and knowledge.
These gatherings were not simply methods for collecting information. They were environments of learning.
People shared stories of family, faith, migration, displacement, technology, neighborhood histories, and everyday practices of care that allow communities to survive and thrive. The most important knowledge was not only contained in the words themselves, but in the relationships created through exchange: recognition, trust, vulnerability, and collective reflection.
The dinners became learning ecologies. They are a form of cultural infrastructure—the mediating spaces where communities decide what is worth remembering and how to pass it forward. The oral histories shared over these meals are not raw material to be scraped; they are the commons from which our collective future is crafted.
They reminded me that humanity has always practiced continual learning. Long before artificial intelligence, humans taught one another how to live through stories, songs, rituals, recipes, gestures, ceremonies, and memory.
AI does not inaugurate learning.
AI enters an ancient tradition.
The difference is that humans are now creating systems capable of learning at unprecedented scale. This makes the conditions under which learning occurs a profound civic responsibility. If learning changes the conditions under which others learn, then teachers do this. Parents do this. Artists do this. Governments do this. Algorithms certainly do this.
So the question becomes: what responsibilities do we have—not only for what we learn, but for the learning environments we create for others?
Data Gifts Create Responsibility
The next transformation moved the stories from language into DNA.
Working with Jeff Nivala and others, excerpts from the oral histories were encoded into synthetic DNA. Because DNA synthesis is costly, only selected excerpts could make this journey. That limitation became part of the work itself. Archives are always shaped by what can be carried, afforded, protected, and passed forward.
The encoded DNA was added to soil where Okra and Black Oak were grown. Okra entered the work as food, ancestor, talisman, and time-traveling vessel—a living technology that traveled with enslaved Africans into the Americas and Caribbean, carrying nourishment,
memory, and strategies for survival. Black Oak became another witness: rooted, slow, ecological, and enduring.
As experimental envoys of community knowledge and care, they asked the question: Can stories enter land? Can memory become stewardship? Can DNA become not only a storage technology but a way of claiming continuity, presence, and belonging?
We were here.
We are here.
Our stories have entered the ground.
This process was built with faculty and students at San José State University, the staff of ICA San José, scientists, artists, technicians, community hosts, and participants. That network matters because a data gift is not merely information. A data gift creates responsibility.
A gift is not free because it is offered. A gift creates a relationship, and relationships require terms: consent, context, limits, reciprocity, stewardship, the possibility of refusal, and the right to ask how something will travel. This stands in direct opposition to the traffic-based value machine that currently governs the internet, which treats all community knowledge as an uncompensated, extractable resource.
Systems of care do not say, “Give me everything and trust me later.” They say, “Slow down, name the terms, and honor that some knowledge is powerful because it is held close.” Care is a deliberate, unending pursuit. It is love made practical and answerable: the ongoing work of noticing what our relationships ask of us; protecting the dignity, agency, privacy, and possibility of those within them; and tending the consequences of what we build.
Care asks not only whether a system works, but whom it works for, what it asks of them, what it carries forward, and whether those caught within it can remain whole. It may arrive as nourishment, listening, and support. It may also take the form of refusal, friction, limits, opacity, redistribution, and repair.
This is not anti-technology. It is pro-relational.
Prismatic reasoning is essential because data is never one thing. A story can be a personal memory, cultural knowledge, training material, artistic source, scientific payload, archival fragment, and sacred offering all at once. Policy often wants to sort these into clean categories. But lived experience does not always arrive clean. Community knowledge rarely travels as a lone object. If data is treated only as a resource, extraction becomes inevitable. If data is understood as a relationship, responsibility becomes unavoidable. But a relational model of data cannot survive without bodies to enforce it. The collapse of our traditional arts organizations and funding models has left an institutional void. We must ask not only who owns this cultural knowledge, but what new collective structures—data trusts, community guilds, or stewardship councils—are required to administer these rights, protect the commons, and ensure that the value generated by AI flows back to the communities that taught it.
This has implications for AI policy. Policy must move beyond regulating outputs and begin governing learning conditions: what systems learn from, who shapes that learning, who benefits, who is harmed, and who has the power to refuse, revise, or redirect it.
A prismatic policy framework would ask not only, “Was the data legally obtained?” but also:
Was the relationship ethical?
Was the context preserved?
Can people refuse future use?
Can communities contest harm?
Does the system return value?
Does it sustain the people it learns from?
Glitch as Survival
When the DNA was decoded, the text was altered
It returned with glitches.
Yet powerful information survived the technological journey. This felt connected to ancestral forms of knowledge transmission. Stories have always traveled through imperfect conditions. They have survived migration, violence, displacement, censorship, and loss because communities cared for them.
Glitch is not a simple failure. It was evidence of passage. Whole sentences arrived carrying memory.
After the text returned, I had it read by a clone of my voice. Hearing that synthetic voice speak the glitchy text was hauntingly beautiful. It was me and not me: echo, double, residue, future ancestor. Hauntology feels useful here—the past refusing to stay past, the not-yet arriving with a familiar breath.
Then an invitation to participate in SHIFT at the Royal Opera House in London opened another possibility.
What if the file did not simply speak?
What if it sang?
The universe sent me mezzo-soprano Samantha Rose Williams. She listened and asked the practical question imagination had not answered: who would write the score? I did not have an answer.
“I guess it will be me and AI,” I said a beat later.
That sentence became a method.
Collaboration Rather Than Replacement
One persistent question surrounding AI is whether it will replace creative work. I think that question misunderstands creativity.
Creativity has never been solely about producing an artifact. It is curiosity, judgment, context, risk, relation, refusal, and the ability to recognize meaning where none was previously visible. Artists have real reasons to question AI: whose labor is hidden, whose knowledge is absorbed, who profits, what gets flattened, and what forms of making are made disposable?
But rejecting these technologies entirely risks losing something important. The question is not whether AI replaces creativity. The question is whether we can create new forms of creativity through careful relationship with emerging systems.
In making the aria, AI did not replace my imagination. It helped me move ideas that lived internally—musical structures, sonic relationships, visual possibilities—into forms that could be examined, questioned, refined, and shared. I tried Claude and ChatGPT. I called on my old high school music knowledge. I imagined a piano, a talking drum, and a mezzo-soprano. I brought in call and response, old Black hymns, Porgy and Bess, Nina Simone, Mahalia Jackson, Audra McDonald, early Zap Mama, and some of Sam’s favorites, including “Mon cœur s’ouvre à ta voix.”
There were many iterations: me and ChatGPT, Sam’s notes, an arranger’s notes, more machine outputs, more human intervention. I used Sibelius, MuseScore, Logic, Ableton, and scripts moving through notation and audio workflows to listen, revise, and test.
I also learned something about myself.
I like thirty-second notes.
The singer, understandably, needed to breathe.
Many AI-generated edits looked plausible on the page but required too much of the human singer. They ignored lungs, vowels, fatigue, phrasing, and dramatic intention. Samantha brought the intelligence of breath. Musicians brought timing, interpretation, and touch. I brought intention, patience, and care through many iterations.
The machine generated possibilities.
Humans created meaning.
But even that phrase needs prismatic pressure. The machine’s possibilities are not neutral. They are shaped by datasets, human labor, cultural patterns, omissions, and inherited biases. To work with AI responsibly, artists must not pretend the system is innocent. We must read the output with side-eyed optimism and imagination. We must ask where the material comes from, what it assumes, what it erases, what it makes too easy, and what human judgment must repair. There was skepticism. AI project piqued the interest of some and became a barrier for others. The mu scians who entered the process expressed fascination even while taking flack from their friends. They argued with the score, humanized it, and eventually gave it the green light. A crew of humans recorded live performances at ONX Studio and Lincoln Center.
No prompt lifted a microphone.
No model drew breath.
No software made the room lean in to listen.
AI did not replace creative work. It multiplied the need for creative judgment. For artists, the task is not to throw the baby out with the bathwater, nor to bathe happily in contaminated water. The task is to test the water, name what poisons it, change its conditions, and decide what can be grown there.
Prismatic reason offers artists a way to stay in the room without surrendering critique. It asks us to engage emerging systems not as saviors or monsters, but as unstable participants in the learning ecologies we are still responsible for shaping.
The Artwork as Learning Ecology
Cosmologies from the Bay is not only an artwork about learning. The artwork itself is a learning ecology.
It brought together community members, artists, scientists, students, institutions, AI systems, DNA technology, plants, musicians, and audiences. Each participant changed the conditions under which the others learned.
Community stories changed the technology.
Technology changes the artistic possibilities.
Musicians interpret the AI-generated score.
Glitch changed our understanding of failure.
The learning process continues through encounters with the work.
The artwork is not a fixed object. It is a temporary ecosystem where different forms of intelligence encounter, challenge, and transform one another.
Prismatic reason is not only a concept. It is the working method. The project does not ask oral history, AI, DNA, opera, soil, and policy to become the same. Instead, they ask what they reveal when held in relation to each other. It refuses the idea that any one system gets the final word.
Critical Thinking as Civic Infrastructure
If learning is relational, critical thinking cannot be treated as an advanced skill reserved for school or professional life. It must be cultivated from birth. We are being born into a world where ground truth is unstable, images can be fabricated, voices cloned, knowledge manipulated, and confidence manufactured by systems that do not understand consequence.
Critical thinking is not cynicism. It is nurturing our now and a tool to craft the future.
It is the habit of asking: Where did this come from?” Who benefits from my believing it? What context is missing? What evidence would change my mind? What is being made invisible? Prismatic reasoning deepens critical thinking by teaching us to look again and again from different angles. It does not accept the first explanation as the only explanation. It asks technical, social, historical, sensory, ethical, and spiritual questions. It understands that “ground truth” has often been defined by those with power, yet it does not collapse into the dangerous claim that all truths are equal.
We need both skepticism and discernment.
We need both openness and standards.
We need both imagination and evidence.
At the individual level, we must practice uncertainty without surrendering judgment. At the community level, we need spaces where people can test information together, build shared memory, and repair trust. At the governmental level, we need policy that protects provenance, consent, privacy, labor, and the right to refuse extraction. At the technological level, systems must be designed to show uncertainty, preserve context, expose manipulation, honor withdrawal, and know when not to act.
This is not merely media literacy. It is civic infrastructure.
Intelligence Should Be Judged by the Futures It Sustains
The dominant question around AI is often: How intelligent is the system?
A better question is: What relationships does this intelligence sustain?
An intelligence that increases efficiency while damaging trust, dignity, ecosystems, or communities cannot be considered successful. Intelligence should be judged by the futures it makes possible.
Does it deepen understanding?
Does it expand participation?
Does it preserve complexity?
Does it support creativity?
Does it help communities remember?
Does it create conditions where more forms of life can thrive?
These are not only artistic questions. They are policy questions.
Prismatic reason pushes us to judge intelligence across more than one axis. Accuracy matters. But so do context, consequence, consent, repair, ecological cost, cultural specificity, and long-term stewardship. A system can be technically impressive and socially reckless. It can be efficient and unjust. It can be generative and extractive. It can produce beauty while depending on hidden harm. To see this clearly, we need many lenses at once.
We must also think ahead to systems that do more than answer prompts—systems that define questions, pursue goals, and solve problems autonomously. Before such systems act in the world, they must be governed by public-interest limits: transparency, auditability, contestability, human recourse, domain restrictions, and the right of communities to refuse deployment. They must know that not every solvable problem should be solved.
That efficiency is not wisdom.
That communities are not abstractions.
That uncertainty requires humility.
That refusal can be protective.
That care is a form of intelligence.
That no system should be trusted with power it can’t share equitably, limit, or relinquish.
Toward Prismatic Policy
If AI policy is to meet this moment, it cannot focus only on outputs. It must address the full ecology of learning.
That means asking who creates the training conditions, who provides the data, who defines success, who absorbs harm, who is invited into governance, and who can say no. It means designing systems with consent, provenance, contestability, attribution, recourse, and community benefit at their core.
A prismatic approach to policy would not treat technology as separate from culture, labor, environment, or history. It would understand that an algorithmic system is also a social system. A dataset is also a memory system. An interface is also a teaching environment. A model is also a container of inherited values. A deployment is also a political act.
This matters because learning does not stop at the machine. Once deployed, systems teach users how to behave. They shape what questions are asked, what answers feel legitimate, what knowledge becomes visible, and whose judgment is deferred to. In that sense, AI systems are not only learners. They are teachers.
So we must ask: what are they teaching us to become?
Policy should require that powerful systems make their learning conditions legible. It should protect the people and communities whose knowledge trains them. It should demand ways to challenge, correct, refuse, or withdraw from harmful uses. It should recognize creative labor, cultural memory, and community-held knowledge as more than raw material. Policy without prismatic reason risks regulating only what is easiest to measure. But the deepest harms are not always easy to measure. Neither are the deepest forms of care.
The Future We Are Teaching Into Being
AI requires us to confront a question that is ultimately about ourselves: how do we become responsible learners?
We must reconcile our misgivings about each other and about technology if we have any hope that technology will serve us well. This does not mean abandoning skepticism. It means recognizing that critique alone cannot build the future. We must nurture technology into care, and to do that, we must nurture one another across racial, political, ecological, species, and technological boundaries.
Before machines can learn to care for us, we must become better at caring for one another. The future of AI-human collaboration will not be determined by computation alone. It will be shaped by the stories, relationships, and responsibilities we build into learning itself.
The story survived because many kinds of intelligence carried it.
The dinner table carried it.
The church carried it.
The community carried it.
The molecule carried it.
The soil carried it.
The OKRA carried it.
The Black Oak carried it.
The glitch carried it.
The machine carried it.
The singer’s breath carried it.
The audience carried it forward.
Prismatic reason asks us to honor every carrier without confusing them. The molecule is not the singer. The soil is not the algorithm. The community is not the dataset. The machine is not the ancestor. But each participates in the story’s journey, and each changes what the story can become.
The question now is whether our machines can learn to carry us—not as data, not as content, not as prediction, but as living, breathing, contradictory beings still becoming. Our stories survived because many kinds of intelligence carried it.
The future depends on what kind of learners we become.
Transparency
Models / Tools Used
yes
Dataset Sources
yes
Consent / Stewardship Plan
yes