Introduction
In S4E4 the popular science fiction TV-series Black Mirror makes its first foray into the future of matchmaking. Titled “Hang the DJ,” the episode tracks the story of Frank and Amy, two love-seekers participating in a dystopian dating program (“The System”) that pairs them with various partners for relatively short periods of time. The System is presumed to analyze the harmony within these artificially assigned relationships and ultimately find a soulmate for each participant. As Frank and Amy experiment with different partners, they develop strong romantic convictions for each other despite repeated interference from The System. When they learn that they have not been paired as soulmates, Frank and Amy choose to be together in an act of rebellion. To the audience’s surprise, it is ultimately revealed that their entire experience was a simulation – one of 1,000 testing the compatibility between Frank and Amy, and one of 998 where they decided to rebel. This purported 99.8% (998/1,000) compatibility is the basis for the pair’s first introduction in base reality at the episode’s conclusion.
With recent advancements in artificial intelligence and virtual reality, can we expect to see a non-fictional version of “Hang the DJ” in our near future? This is a resounding no. The level of sophistication required to create and manage thousands of intricate simulations, each mimicking complex human behaviors and interactions, is beyond our current technological capabilities. Such simulations would demand vast computational power and an in-depth understanding of human psychology beyond our capability. Simply put, we are nowhere close to being able to transfer human consciousness into simulation.
With this being said, the impossibility of simulation-based matchmaking does not imply that any attempt to utilize technology to measure compatibility is necessarily futile; after all, algorithmic matching programs grounded in psychological research have been used since the 1960s (1). In what follows I will briefly discuss modern-day services that perhaps resemble “Hang the DJ” and disagree with notion that they are dystopian.
Snack & Checkmate
Released in 2021, the Snack dating app represents a bold re-imagination of what online dating can be. Amongst other unconventional features, Snack allows users to create AI-trained avatars that other users can interact with. The ultimate goal is to create for yourself a “digital doppelgänger” that can converse with, and screen users by itself. The avatar is made to physically resemble you and its “personality” trained through a series of interest selections and text-based Q&As. Active users can then explore a map populated by AI avatars and chat with them. It is like talking to a LLM (e.g, chat GPT) that is trained to act like the user it represents (sort of but not really). As a concept, Snack’s similarity to “Hang the DJ” is striking; the platform is quite literally striving to simulate your romantic persona. In execution, however, Snack leaves quite a bit to be desired.
Checkmate is a recently released app from the creators of Marriage Pact that allows any two individuals to approximate their long-term compatibility. Upon signup, a user is required to answer 24 personality-based questions that will ultimately be interpreted by Checkmate’s algorithm to calculate their compatibility with someone else; you can add friends within the platform and/or scan one’s QR code to initiate the process. Similar to Snack, users are able to answer additional questions that over time strengthen their personality profile; antecedent compatibility ratings are readjusted in real-time. The app exploded in popularity in late 2022 but has appeared to quiet down since. Marriage Pact also has plans to release a complementary service, “Soulmate Radar,” which will notify two unaffiliated users of their high compatibility when in close proximity. Taken together the services might be thought to resemble “Hang the DJ” by using compatibility metrics to facilitate impromptu real-world interactions.
“A Dystopian Nightmare” – Chunhua Yu on Checkmate
In an opinion article published by the Stanford Daily in April 2023, Chunhua Yu warns us of the dangers of “Hang the DJ” replicas (2). Titled “Checkmate and AI: Reality is becoming dystopian, but we might not know it,” Yu specifically criticizes the aforementioned Checkmate for seducing its users into evaluating their existing relationships based on overly-simplified compatibility metrics. Presumably Yu charges Checkmate’s algorithm for a “lack of depth” based on its inability to accurately reflect the perceived compatibility of her existing relationships; she reports having received a mere 2% compatibility rating with a close friend and never any high scores. She further comments on the irrationality of the Checkmate craze: on one hand users understood the “misleading” nature of the app but were still all too eager to compare, analyze, and find legitimate explanations for Checkmate’s ratings. In a word, Yu concludes that Checkmate takes advantage of our natural desire for self-knowledge using groundless numerical shortcuts that contradict our actual intuitions. She is concerned that we have a natural propensity to ignore these intuitions in favor of grandiose claims regarding the algorithmic explainability of subjective phenomena.
Coming from a background in philosophy I cannot resist the temptation to restructure her argument in “premise-conclusion” form. It would look something like this:
Premise 1: Checkmate’s “simple, misleading” compatibility ratings are inaccurate.
Premise 2: Even so, users are still tempted to “compare,” “make sense of them,” and form “conclusions about who we are and what our relationships look like”
Premise 3: These activities promote feelings of jealousy and self-doubt.
Conclusion: Checkmate is bad, alarming, a nightmare, etc.
Response:
In my view, any substantive criticism articulated in Yu’s article is unfortunately overshadowed by a sensationalized perspective on AI ethics. “Dystopian” in this context is a misnomer. People also read too much into, and are entertained by astrology signs, IQ tests, Rice Purity scores, and the Myers-Briggs Type Indicator. I have never heard someone refer to these as dystopian. Why should the introduction of machine learning make Checkmate so? Indeed, Yu seems to think that Checkmate does no better of a job at quantifying subjective traits anyways. The problem for her, then, is that other people want to believe they do. For one thing, I think it is a stretch to label that as a “nightmare.” Furthermore, already baked into this desire to believe is the notion that people do not actually hold Checkmate’s ratings in higher regard than their own intuitions; otherwise the pervasive skepticism she speaks of would not exist. Indeed, perhaps the best aspect of Checkmate is that we can question its findings and hold it accountable. On the contrary, sites like eHarmony rate compatibility exclusively amongst strangers that have no way of actually “gut checking” their scores. In summary, it would appear that Yu’s argument is primarily a critique of Checkmate’s current accuracy. Needs improvement? Sure. A dystopian nightmare? No.
What if Checkmate was error-free though? What if it always gave compatibility ratings that were perfectly in line with our existing intuitions, and accordingly made accurate compatibility predictions for those we do not know? Would Yu still condemn Checkmate as dystopian? I hope not. Such a service would be a fantastic tool one could use to meet new people guaranteed to ultimately be great friends. And from an ethics perspective, that is what AI should be: a tool used by self-determining individuals to help them make better decisions for themselves.
Contrary to Yu, I think we enter the realm of dystopia when powerful institutions are using scientifically dubious metrics to make decisions for, or on behalf of people. IQ tests in a vacuum? Fine, interpret them however you would like. Governments using these scores for socioeconomic placement? Not fine. Going full circle, the real dystopian element in “Hang the DJ” is not when real-world Frank and Amy match at the end, it is the world their simulated-selves are condemned to: every waking minute of their days is controlled by some menacing “System” that justifies its actions through the frequently repeated mantra: “there will be a reason.” Luckily, and contra Yu, the System is not “already here” and we do not have to worry about Checkmate downloading sentience any time soon.
Sources
(1) Slater, D. (2013). Love in the time of algorithms: What technology does to meeting and mating. Current/Penguin Group.
(2) Yu, Chunhua. “Checkmate and AI: Reality Is Becoming Dystopian, But We Might Not Know It.” The Stanford Daily, 2023.