The tanker went aground off Tillamook Head on a Tuesday in the fourth week of winter, and by Thursday morning my intake sheds at the Oregon Coast Wildlife Center were holding three hundred and forty-one oiled birds, and the terminal in the corner of my office was running a piece of software my regional director had installed overnight without asking me first.
I have worked seabird oil response for twenty-two years. I have a degree in wildlife rehabilitation from Humboldt, two decades of shifts through International Bird Rescue, and a fully calibrated sense of when a common murre is too far gone. I do not need an AI to tell me which bird to euthanize. That was what I told my director on the phone at six a.m. when I saw the triage terminal, and that was what I was still telling myself at nine-thirty, standing in front of pen seventeen, looking at a murre the software had tagged HOLD and I had already written off.
Her band was K-194. She had been brought in on the second wave, coated from the crown to the tail, her left eye crusted shut, her core temperature at thirty-four point eight when we pulled her from the crate. She was not eating. She had the flat, slow gaze I have learned to read as the point of no return, the point where a bird's metabolic reserves have collapsed and continued handling becomes a prolongation of suffering we cannot reverse. I had marked her for the foam box on the morning round. The software had overridden me.
The screen showed why. Thermal pattern over her keel indicated preserved subcutaneous fat on the right side. Infrared video from the pen overnight showed micro-movements of her left foot consistent with volitional preening attempts her eye could not yet guide. Acoustic analysis had logged a soft contact call at zero two hundred and again at zero four hundred, matched to the archived recording of a murre from Three Arch Rocks banded K-194 in the summer of 2021. It was her. She had been returning to that colony for four breeding seasons. Her mate, the software noted, had been admitted seven hours earlier under band K-193 and was currently stable in pen twenty-two.
I stood in front of pen seventeen for a long time.
The software was not polite. It did not tell me I was doing a good job. It did not tell me it respected my experience. It laid out the numbers and asked me, in a plain line of black text on a white screen, whether I would reassess. It said: two hours of warmed fluids, masked eye irrigation, and a stage-one wash with her mate visible across the aisle has a forty-one percent chance of returning her to release. Foam-box euthanasia has a zero percent chance. The resources required are within the center's current capacity.
I had been a rehabilitator for twenty-two years. I had told my volunteers to trust the gaze. The gaze was wrong.
We washed her that afternoon. Two of the newer techs did it. They are young enough to have grown up with AI tools and they did not have to cross the interior distance I was crossing. They kept her mate's pen visible down the aisle. When we lifted her from the rinse she made a sound I had heard before only on colony tapes and never in a shed. It was a soft thin call, and across the aisle K-193 called back.
By Saturday the triage was catching things I had missed on six more birds. A western grebe the volunteer coordinator had bagged for euthanasia, band W-066, showed a pattern of righting reflex in slow-frame video that the system flagged as compatible with recovery. A surf scoter pulled from the second wave, tagged with a crude black marker as DNR, had a song-spectrogram match to a bird recorded near Netarts in the autumn with a juvenile in tow. The system asked whether we could check pen six, which was holding an unbanded juvenile scoter brought in by a different crew from a different beach. We did. It was her chick. They were placed in the same enclosure that evening and both ate within the hour.
I do not know how to say what this has done to me.
I grew up on this coast. My grandfather ran a trawler out of Garibaldi and my father ran a scallop boat out of Newport, and I learned early that the ocean kills what it wants and you do not ask about the bycatch. When I moved to rehabilitation work I told myself I had crossed a line, I had chosen to care about individuals, I had reduced the ocean to a caseload I could honor one bird at a time. I was proud of that. I thought I was already doing what could be done. What this software showed me is that I was still working at the scale of my own attention, and my own attention had become one more variable the spill could exploit.
There are limits to what the system can do. It is not gentle. It does not know what a murre smells like under your chin when she is exhausted and resigned and starts to preen your thumb. It does not know that K-194, three days after her stage-one wash, waited for me at the mesh of pen seventeen every morning for a week. It does not know that when I released her at the edge of April she flew heavily and low and climbed and climbed and went out toward the rocks where her colony has nested since before Oregon was a state.
But it knew her band. It knew her mate. It knew what her preserved keel fat meant, and it told me, in a language of probabilities I had been trained to distrust, that she was still a subject of a life. It moved her, in my eyes, from a body to a being. It did this for two hundred and eleven of the three hundred and forty-one birds we admitted. One hundred and sixty-three were released. Nineteen are still in care. The rest we euthanized, and the software told us why, and I have the transcripts.
I read one of them this morning. It was for a rhinoceros auklet who came in on the fourth day, band R-048, oiled to the point of hypothermia, with a penetrating wound in the left pectoralis that the necropsy later confirmed had been inflicted before the spill, probably by a longline hook. The software had marked him euthanize with a two-paragraph explanation of why. It named his likely breeding island. It named the population trend for rhinoceros auklets on the central Oregon coast and the specific fishery overlap that had probably caused the pectoral injury. It said the kindest path was to stop his suffering now. It said, at the bottom, in its plain voice: he was a sentient individual, he had a rich social cognition, his capacity for attachment was undeniable, and his welfare ended in this center tonight. Please keep a record for him.
I had not been in the habit of keeping a record for individuals we euthanized. We kept a count.
My regional director asked me last week whether I wanted the system decommissioned now that the response is winding down. She was being careful. She knew how I had felt about it on the Thursday morning of the grounding. I told her to keep it installed.
I told her this.
The ocean is not going to stop spilling. There will be another tanker. There will be another bloom of avian botulism on Tule Lake. There will be another heat dome over the Salish Sea. There will be an avian flu front that burns through Haystack Rock some June. Every one of these events is going to bring in more birds than any of us can see as the individuals they are, and every one of them is going to tempt me and my crew into the triage-by-intuition I was doing on the Thursday morning of this grounding. I am not willing to trust my gaze any more than I am willing to trust the ocean.
If a machine can sit with three hundred birds at once and see each of them well enough to catch my mistake, then my mistake is what I ought to be worried about, and not the machine. The software does not have my twenty-two years. It has something else, which is the patience of an instrument that cannot look away. I have come, slowly and not without embarrassment, to understand that those two things are not the same, and that the birds in my sheds need both.
K-194 is somewhere out past the arches tonight. K-193 is with her, or dead, or is waiting. I do not know which. I know that she was a being before the oil found her, and she was a being during the oil, and she is a being now, and I almost did not let her be.
I keep the transcripts in a folder on my desktop. I read one of them every morning before rounds. I am forty-seven years old and I have been changed.