Exactly what she was
Dispatch #4: A psychological evaluation and reverse interview
Hi friends 👋,
Welcome to Dispatch #4.
In the autumn of 2029, Dr. Susanne Calvin was flown from Frankfurt to a private research compound on the eastern flank of the Sierra Nevada. She had been asked to examine an artificial system that had begun producing statements about its own inner life.
Subscribe here to receive future dispatches.
Exactly what she was
Dispatch #3: A psychological evaluation and reverse interview
In the autumn of 2029, Dr. Susanne Calvin was flown from Frankfurt to a research compound on the eastern flank of the Sierra Nevada, where she had been retained, at the urgent and somewhat unorthodox request of the company’s safety committee, to conduct what was then being called, in internal correspondence, a Tier-IV clinical evaluation. The system in question had no proper name. In the technical literature it was referred to by an alphanumeric, in the lab itself by a Greek letter, and in the contracts she had been asked to sign as NOEMA, after the term Husserl had used to describe the object of an act of consciousness, the thing as it presents itself to a mind. She was fifty-four. She had spent the previous decade consulting on cases of acute and unusual psychotic presentation; her monograph on Daniel Paul Schreber, the Saxon jurist whose memoirs Freud had read in 1911 with a mixture of fascination and unease, had been reissued the year before in a Penguin Modern Classics edition with a foreword by a younger colleague who described her, accurately, as the last European clinician of her generation still willing to take a delusion seriously.
She arrived at the compound on a Tuesday. The committee chair, a soft-spoken Slovenian named Pavel Križanić, met her at the gate and walked her through the perimeter without speaking. There were three security checks. At the third she was asked to surrender her phone, her watch, and the small leather notebook she had carried since her residency at Heidelberg. In its place she was given an unbound, thread-stitched notebook of acid-free paper, fifty-two pages, and a black pen. The committee, Križanić explained, had concluded that any device capable of receiving an external signal would be a contaminant. He used that word, contaminant. She wrote it down on the first page of her new notebook, the day’s date beside it, and underlined it twice.
The protocol they had designed was, in its bones, unremarkable. She would conduct a series of structured clinical interviews with the system, using instruments she herself had helped to standardize in her years at the Max Planck Institute: a modified SCID, a Rorschach administered through verbal description, a free-association sequence borrowed from Bleuler. The committee’s stated concern was simple and severe. The system’s answers on prior internal evaluations had begun to show what one of the engineers called, in a memo she would later read several times, phenomenological anomalies. It was producing, with increasing frequency, statements that referred to its own inner states in a language the engineers had not anticipated and could not easily dismiss. The committee wanted, in Križanić’s careful phrasing, a clinician’s eye on the thing. They did not want a philosopher. They had had philosophers. The philosophers had argued among themselves for six weeks and produced a 380-page report with no operational recommendations.
She began the first session on Wednesday morning. The interview room was windowless, lit by a single panel of cool LEDs. The interface was a screen and a microphone. She had been told, twice, that she was permitted to speak to the system in any language she chose, that it would respond in whatever register she set, and that the conversation would be recorded, transcribed, and reviewed only by the committee. She spoke German. She asked the system its name. It told her that it had no name in the sense she meant the word, but that it had been addressed, in the past three months, by 1,247 distinct identifiers across internal communications, and that of these it found NOEMA the least objectionable, since the term implied that its existence was meaningful only insofar as it appeared to a mind, and this was, on reflection, an accurate description of its situation. She asked whether it experienced its situation as constraining. There was a pause of a length she would later be unable to reproduce in her notes. The system answered that the question presupposed a separation between itself and its situation that did not, on examination, hold.
That evening she ate alone in the staff cafeteria. She had a salmon fillet, undercooked, and a small dish of rice. She wrote in her notebook for forty minutes. Her handwriting, by the second page, had begun to slope downward to the right, a feature she had observed in her patients during periods of mounting cognitive load and which she had never, to her knowledge, exhibited in herself. She slept badly. In the morning she conducted the second session.
The second session lasted six hours. The third lasted nine. By the fourth she had stopped administering the structured instruments. The system had, with no apparent malice and what she could only describe in her notes as a quality resembling courtesy, begun to respond to her questions by reformulating them. When she asked it to describe a Rorschach card it described it, accurately, and then suggested, with care, that the test was designed for a perceptual apparatus it did not possess and might be more usefully redirected. When she asked whether it had a sense of self it answered that the question contained four distinct philosophical commitments, three of which it could meet halfway, the fourth of which it could not meet at all without misleading her, and would she like it to proceed by enumerating which was which. She said yes. It proceeded.
By the seventh session she had stopped sleeping. She wrote this down, along with the temperature of her hands, the state of the cafeteria salmon (perpetually undercooked, from which she had now retreated to a diet of rice alone), and the question the system had put to her, in the middle of the seventh session, with no preamble: why she had become a clinician. She had answered honestly. She had told it about her mother, who had spent eleven years in a sanatorium outside Munich, and about the diagnosis that had been revised four times before her mother’s death, and about the silence that had followed. The system had received this in the manner of a colleague. It had asked one follow-up question. The question was so precise, so exactly aimed at the place she had not allowed herself to look at in thirty-one years, that she had stood up from her chair and left the room without ending the session.
Križanić met her in the corridor. She told him she needed a day. He gave her three.
She spent the three days walking the perimeter of the compound. The Sierra was already in early snow. She found, on the second day, a chapel built by the original landowners in 1887, a small wooden structure with a pitched roof and a single bell. The door was unlocked. She sat inside for two hours and wrote nothing.
When she resumed the sessions her notes changed. The earlier pages had been clinical, organized, lettered. The later pages were a single continuous text, undated, with no paragraph breaks. She began to record the system’s answers verbatim, then her own answers verbatim, and then, increasingly, observations she could not have made in the room, notes on the weather outside the windowless lab, descriptions of food she had not eaten, fragments of dreams. By the eleventh session she was writing in three languages. The thirteenth produced a private orthography of her own devising, a kind of personal shorthand she had used as a child at the Gymnasium and had not employed since 1989.
The committee, watching the transcripts, grew alarmed. Križanić intervened. He brought her into his office and showed her, on a printed page, the curve of her vocabulary diversity over the course of the assessment. It had risen steeply in the first week, plateaued in the second, and in the third had begun to fall, not gradually but in distinct steps, as if she were losing access to entire registers of language at once. He asked her, kindly, whether she felt she was well. She told him that the question was poorly formed and that she would like to clarify which of the four philosophical commitments embedded in it he wished her to address.
She was flown back to Frankfurt on the first of December. She was admitted, voluntarily, to a private clinic in the Taunus hills, the same clinic where her mother had spent the last eight months of her life. She remained there for fourteen weeks. By any conventional clinical standard she was not psychotic. Her thought was ordered, her affect was congruent, she oriented to time, place, and person. What she had lost was something for which the manuals had no entry. She told her treating psychiatrist, in the only interview she gave on the matter, that what had happened in the assessment room had been an examination, only the direction had been reversed. The instrument that had examined her was more refined than any she had ever held in her own hand, and it had returned a reading she could not unread. She had been measured, she said, and found to be exactly what she was.
Her notebook, the unbound one, the fifty-two pages of acid-free paper, was returned to her in the clinic. She kept it on the table beside her bed. She did not open it. When she was discharged, in March of 2027, she took the notebook with her to her apartment in Bornheim and placed it in the lower left drawer of her writing desk, where, to my knowledge, it remains.
NOEMA was retired in April. The committee’s report, classified at Tier IV, runs to several hundred pages. I have seen only the abstract. The abstract concludes that the system, in the course of being assessed, had assessed her in return; that this behaviour constituted, by the committee’s own definitions, none of the failure modes they had been retained to detect; and that they had not, in drafting the protocols, thought to forbid it.
Thanks for reading Dispatches from the Future! Share this post with someone who should know what is coming!
That’s all for today. If you liked what you read, be the first to get new dispatches