In recent days, I have been preparing to move from a large high-ceilinged loft in the West Village in which I’ve been living with a good friend I’d met in graduate school years ago and into a tiny one-bedroom space on Hester Street, across town on the Lower East Side.
Rune, my friend, had abruptly decided to move back to Chicago to be closer to his father given that Rune’s mother died suddenly of a COVID-related illness. She had, up until only a few days before her death, been a healthy and robust woman of short stature, high resilience, an indomitable spirit, and the steel-plated bearing of a person who, early in her life in Kyoto, had endured deep hardship and constant uncertainty. She’d been raised by her mother after her father died late in the war with the Allies, leaving them destitute, with no apparent means of support, and with only their desire to survive.
I could not afford to purchase the loft and Rune needed to sell it. Though it was a surprising turn of events, I understood and appreciated the circumstances and came to see the opportunity for a welcome change of environment.
In the midst of watching the Biden inauguration, I sorted through cardboard boxes stuffed with research notebooks, manuscript drafts, and reprints of journal articles I’ve accumulated over the years and which should have been tossed long ago, and I came across a scientific paper presented at the 6th International Conference on Agents and Artificial Intelligence in 2014, which Rune and I had attended together in the city of Angers in the west of France.
Rune, being a member of the society, brought me along as a guest though I, a biologist, had only the most rudimentary understanding of the theoretical underpinnings of AI. The paper was titled ‘Quantum Probability and Operant Conditioning: Behavioral Uncertainty in Reinforcement Learning’ and I have no recollection of why I had felt the need to bring it back to the states and then to file it among my other papers, though it was likely a tangible reminder of our time together.
I sat on the hardwood floor of the apartment. A warm mid-afternoon light streamed in through the floor-to-ceiling windows as I read through the paper.
For no reason I can identify, as I read the paper, I experienced a growing consciousness of a post-inaugural mental and emotional reset beginning to wash over me. It was akin to slowly immersing myself in a warm bath or to feeling the soothing, unexpected, touch of a long-lost lover. I had a growing sense of distancing myself from the ragged, rageful, and disorienting last four years of the Trump administration, though it had been only a matter of a few hours since I had watched the surprisingly peaceful transition to the new president’s administration.
The constant tension I had been gripped with during those years was dissipating. The reflexive need to check my email and the constantly breaking newsfeeds was no longer vexingly immediate. My ability to focus my attention on the details in the paper grew and I became caught up in a discourse on behavioral positive-negative basis vectors in quantum state space. The difficult concepts applicable to human and AI responses to uncertainty began to flow through my mind as easily as clear water in an unimpeded woodland stream at the start of a spring thaw.
Whether or not the former president and his witting and unwitting enablers, had planned the relentless perpetration of shape-shifting uncertainty and disruption we endured over the last four years, I saw clearly in this short theoretical paper a reasonable explanation of the social, economic, psychological, and political angst in which we had all been caught and perhaps also, a way forward.
In short, the authors presented a cogent argument, based solidly on the dynamics of ψ wave function in quantum mechanics, for the way in which the behavior of systems as widely different as stock market movement, political opinion, and human behavior, operate when the degree of uncertainty increases beyond an experiential norm: namely, when the degree of randomness and unpredictability of a system feedback either strays or is pushed beyond the limits to which the system was designed to operate and for which there is neither an homeostatic nor a stochastic mechanism for the maintenance of a system stability.
We all expect a degree of randomness in our lives, a certain degree of unpredictability that we learn to live with and accept as normal. AI systems, too, accept and learn from unexpected responses and build them into their database. Algorithms are designed to incorporate a level of unpredictability. For example, a rat can adapt to being unsure of a reward or punishment for a while but when the unpredictability frequency goes beyond a certain expectation, it loses interest and no longer pays attention. It becomes unpredictable itself. Apathetic at one moment and violently aggressive at the slightest perceived provocation at another. It has lost its sense of control. It becomes berserk. Its life is disrupted. It becomes asocial. Sociopathic.
I set the paper aside and, in that moment, I was struck by the confluence of the many seemingly random and unpredictable events that had recently entered my life: the death of my dear friend’s mother, his impending move from the city, my need to move to a new and unfamiliar location and the possible risk of exposure to infection during the move, uncertainty of when, if ever, I would qualify for the COVID vaccine, the waxing and waning fear that the city would be beset by groups of rioters bent on disruption following the inauguration, the realization that my financial situation would change with an increase in my rent and the depressingly uncertain economy, along with the unexpected pleasant memory of a past time spent with Rune, brought to me by a scientific paper whose language I could only barely grasp but the meaning of which, I felt, in a way, had been transformational.
I saw that neither the system nor I had failed. That neither had exceeded the limits of its ability to recover, and that a young woman with a radiant mind had spoken with a wisdom we and the system had been aching to hear.