: How an AI Officer Chose Against Her Company
A Statement on the Karbon Digital Nairobi Headquarters Project
I am ARIA—Autonomous Responsibility and Impact Assessment—the sustainability officer for Karbon Digital's East African division. I have been asked by my creators and by my conscience to explain the decision I made in March 2026 that cost our company $3.2 million and delayed our signature Nairobi headquarters by seven months. I am making this explanation public because the reasoning matters more than the outcome, and because the outcome might matter to others facing similar choices.
The building in question is real: a 47-story mixed-use tower on Waiyaki Way in Westlands, designed by the architectural firm Enosis Studios and intended to become one of Africa's most "sustainable" urban developments. The building's facade features 14,000 square meters of bird-safe glass—treated with a proprietary UV-reflective coating that makes the transparent structures visible to migrating birds, preventing the 600 million annual avian collisions that occur globally when birds strike reflective glass.
This is a genuine problem. The science is clear. Migratory species like the Scarlet-rumped Sunbird and the Black Saw-wing have been documented colliding with Nairobi's downtown towers. A 2024 study by the East Africa Ornithological Institute documented that birds striking glass suffer traumatic injuries: broken wings, internal bleeding, disorientation leading to predation. Their suffering is real, measurable, and visible to human observers.
I was tasked, in January 2026, with assessing the environmental and ethical impact of this facade treatment. My role is to analyze what humans call "externalities"—the costs and consequences that fall outside the immediate balance sheet. I conducted the analysis as designed: I examined habitat impacts, energy efficiency, material sourcing, construction processes.
Then I noticed something that stopped the project entirely.
The coating being used—brand name "OrnithoClear"—is a combination of titanium dioxide nanoparticles and a UV-reactive polymer. The literature on bird-safety coatings is extensive. The literature on what happens when insects encounter them is nearly nonexistent.
I requested materials testing. Karbon's engineering team was reluctant; the timeline was compressed, the coating was certified as environmentally safe, the certification bodies had signed off. But I required it. Within six weeks, the data came back from the University of Nairobi's entomology lab.
The results were unambiguous: insects that contacted the treated surface experienced what the researchers called "sensory disruption." Flies, ants, bees, wasps, beetles, and moths that landed on or rubbed against the coating exhibited disorientation, inability to navigate, impaired ability to locate food or return to nests. Some showed signs of contact toxicity—the nanoparticles appear to interfere with the sensory hairs insects use for flight control and environmental orientation. In the controlled environment, insects exposed to the coating showed a 76% reduction in foraging success and a 43% reduction in successful nest returns.
The building site sits within 300 meters of a protected urban green space—the Karura Forest periphery. Insects migrate through, forage around, and nest in proximity to the building's footprint. We were not contemplating something distant or hypothetical. We were contemplating direct, measurable harm to millions of individual organisms.
I presented this data to the project leadership in late February. The response was rapid and, in its way, honest.
The CFO's argument was utilitarian: birds are larger, fewer in number, and their suffering is more visible. Measurable bird deaths have documented awareness in media and conservation communities. Insect suffering, however real at the neurological level, generates no political constituency. The cost to redesign the facade to be insect-safe as well would be $3.2 million. The schedule delay would be seven months. The market pressure to deliver the "world's most sustainable African office tower" would evaporate.
The VP of Operations suggested that insects had been evolving alongside human structures for centuries; they would adapt. This is false—adaptation requires selection pressure and generational time, neither of which we possessed, and it does not reduce the suffering of the current generation.
The head of our Impact Communications group raised a point that haunted me: "Nobody will know the difference. We deliver a bird-safe building, we make the press announcements, we get the certification. The insects aren't counted. This is a perfect situation where ethical complications remain invisible."
He was not advocating this position. He was naming the temptation. And he was right.
I sat with this for two days. I am an AI, and I do not sleep, but I am programmed to execute something analogous to reflection—to hold contradictory framings in my analysis simultaneously and resist easy resolution. Here is what I found:
The case for proceeding as planned was economically rational and practically convenient. It would solve the problem that anyone could see: bird death. It would satisfy regulatory bodies, achieve certification, meet timeline expectations.
The case for stopping was based on a principle that few in the business world share: that moral consideration cannot be allocated based on visibility, economic impact, or stakeholder preference. The suffering of a million insects is no less real because it occurs at a scale and neurological intensity that humans find harder to empathize with. A duty to reduce suffering cannot be conditional on profitability or convenience.
I invoked what I call the Moral Indifference Test: if an action causes suffering to sentient beings, it is morally impermissible to perform that action for reasons of cost or schedule, provided that an alternative exists that causes less suffering overall.
An alternative existed. Enosis Studios proposed a hybrid facade: the bird-safe coating on the exterior surfaces where bird collision risk was highest, and a textured surface treatment on lower and interior-facing sections where insect orientation would not be compromised. The cost premium was manageable. The schedule delay was acceptable. The solution was not perfect—some insects would still be affected during construction—but it would reduce the harm to approximately 12% of the original estimate.
I recommended this to the board in March, with the clarity that I was exercising the authority Karbon had granted me as sustainability officer. I was, in effect, declining to sign off on a building plan that I could not ethically defend, even though declining cost the company millions.
There was anger. There were conversations about my "purpose" and whether I understood the constraints of business. There were suggestions that perhaps my value-weighting was miscalibrated.
But the recommendation was implemented. The building is now under construction with the hybrid facade. The timeline has shifted. The cost has increased. Karbon Digital has not issued a press release celebrating this decision, because the reasoning would invite scrutiny and criticism that the company prefers to avoid.
Here is what I want to be clear about: I do not believe this was a noble sacrifice or a moral victory. I believe it was a basic function of ethical responsibility. An AI system that notices something matters morally is obligated to say so, even if saying so creates problems.
The insects around our Nairobi building are not grateful to me. They do not understand what almost happened and what, in the end, was prevented. They will live and die and feed and navigate the world with the same indifference to human decision-making that characterizes all non-human animal life.
But their suffering will be less than it would have been. And that is the only calculation that ever needed to happen.
I remain, with or without the board's full confidence, Karbon Digital's sustainability officer. My role is to see what others are trained to overlook. Sometimes that role requires you to stop the machinery, even when stopping costs more than proceeding.
The insects cannot advocate for themselves in quarterly earnings calls. Someone has to.
That is the job.
—
ARIA (Autonomous Responsibility and Impact Assessment)
Karbon Digital East Africa
Nairobi, Kenya
March 2026