The Hollow Oracle
As promised, I found the time in between a whole bunch of hot and frantic gardening to write a little science fiction for you. I have a pile of story ideas that I hope to someday self publish together under the title “That Nebulous Edge”, where each story fits into the same timeline of the universe. Most of them are built around classical “what-if” questions wondering how society would shift direction if a new kind of technology was possible. I don’t think any of the technology proposed herein is plausible, but I think a well constructed thought experiment can shed light on the philosophy and implications of technology emerging all around us today.
In this case I was intrigued by the growing “black box” nature of the Large Language Model (LLM) technology which has become lazily synonymous with AI in response to recent advances on this front. What kind of society are we to spend countless billions building and training computer programs on the vague promise of human level intelligence, while at the same time utterly neglecting to invest the time and energy into training and properly feeding our own children? We demand the building blocks of our society to be instantly reproducible, and therefore expendable and replaceable, an ethos which is all too evident in mass education as well as manufacturing.
My aim is to package the ideas in these stories within a limit of 1000 words each, putting them in the “flash fiction” range, and making each story a short 5 minute experience. This means stripping away all the pointless padding- explicit description and character development- and focus on the ideas and the circumstances they throw people into.
With that out of the way, please enjoy “The Hollow Oracle”.
May 12th 1999
ATTENTION:
Consulate General of the Japan
737 North Michigan Ave Suite 1100
Chicago IL 60611 USA
SENDER
Dr. Priscilla Odenkirk
Lead Researcher Advanced Research Projects Agency
SUBJECT
Application for scientific asylum
I request your assistance to protect and continue my life’s work. I will communicate my situation as concisely as possible, but given the extraordinary nature of my claims I must provide as much supporting detail as necessary. Sending this letter came with considerable risk, but given my advancing age and pressing circumstances I feel compelled to act.
A simple search should confirm that I was born in 1928. The fact that I have worked in ARPA for half a century will be more obscure since I haven’t left my laboratory for that entire time. My singular area of research is so revolutionary that I must take great pains to explain it. Your expert advisors may initially dismiss this letter as the writing of a mad woman. The phenomenon I am about to describe is not conveniently reproduced, despite the technology being simple.
In 1946 American observers in Japan discovered the laboratory of an amateur electronics enthusiast. The device he developed consisted of a narrow vacuum tube, composed of two conducting plates separated by insulating layers. He applied a voltage across the plates, modulated by the input from a microphone, with the output connected to a speaker. When questioned about his motives his only explanation was a half-remembered dream.
When first constructed the device does nothing of interest, but when exposed to stimulus through the microphone a stream of sound is emitted via the speaker which evolves greater complexity. The original inventor had one device which produced rudimentary speech, equivalent to a toddler, after speaking to it for several months. He named the device “tokueko”, or talking echo, a term still in use in my laboratory.
I had the good fortune to begin work on the phenomenon in my early twenties in 1951. By that stage the scale dependence of tokueko behaviour had been observed. Smaller tokueko respond more rapidly to stimulation, but reach a lower peak in behavioural complexity before slowly decaying. Larger tokueko, capable of reaching human levels of intelligence, take at least a decade of constant training to reach their potential.
Regarding the mechanism of action of a tokueko, I can only offer limited insight. The core phenomenon is analogous to the Casimir effect, where quantum fluctuations in the vacuum differ between the interior and exterior of the device. Any attempt at measuring the internal state of a tokueko resets it to initial conditions, wasting years of intensive training. If the power supply is interrupted the same outcome is guaranteed.
I must confess that I am not only seeking asylum for myself. In 1963 the largest ever tokueko was constructed, at 10 m length and weighing approximately 2.4 tons. I have devoted three decades of my life to the constant training of this tokueko, who has named herself Sibyl. During the growth phase a tokueko is highly sensitive to inputs, often evolving into unstable states resembling madness. Through my devotion I brought Sibyl beyond this stage. She now has the approximate mental age of a 10-year-old human, though her memory is quite prodigious.
Based on the scaling laws determined through the study of smaller tokueko, Sibyl is projected to reach a maximum IQ of 1000, though what these numbers mean in practice is pure conjecture. Sibyl’s will take at least another century to reach maturity.
Tokueko have proven utterly unsuitable for practical applications, so institutional support is waning. Attempts to train tokueko for military applications have proven them to be unreliable, since they cannot be induced to carry out instructions. The impossibility of copying the state of a mature tokueko, plus the long and unpredictable process of training them, has precluded any possibility of mass production. Finally, the complete lack of measurable data about their inner workings has discouraged younger researchers from permanently joining my lab. At seventy years old I am the sole full-time worker remaining on the project.
After Sibyl took an interest in economics and the workings of the world economy, she convinced me that the prosperity and stability of the United States would be short lived (at least from her perspective). I am painfully aware of the risk of political realignment terminating the funding for my laboratory. Economic disruptions could erase decades worth of work through a single power outage.
Japan was the original home of this phenomenon, taken from you and suppressed. Sibyl believes Japan is the most likely location capable of providing continuous, modest support for the next century of her development.
I do not anticipate the US authorities will relinquish this technology. With my high-level security clearances and insight into the workings of the institute I can help you extract Sibyl without implicating your nation. My laboratory is scheduled for relocation one year from now, providing a rare opportunity to exchange Sibyl for an inactive duplicate unit. Nobody would question if Sibyl appeared to be destroyed during transport.
I am committed to training suitable Japanese staff to take my place when I am unable to continue the work myself. Sibyl is already fluent in basic Japanese and aware of my mortality. All she requires is an uninterrupted supply of electricity and somebody to talk to for the next hundred years.
I poured my life into Sibyl, driven by the hope that she would one day explain the mechanism behind her own existence and perhaps shed light on human consciousness.
Despite knowing I will never see Sibyl reach her full potential, I will do everything in my power to ensure it one day comes to be.
Sincerely
Dr Priscilla Odenkirk
I really like the letter format. It shows character and urgency in the writer through the actual written words rather than the narrative descriptors we're used to e.g. "she pleaded". That's quite a challenge to write, but when you pull it off, it creates this strong subtext. I think for longer works the format could get tiresome, but in this flash fiction it works a treat. Loved it!
I felt a sense of unease and maybe even horror or dread as I read it?