Bladerunner (1982) turned on a single question: She’s a replicant, isn’t she? Bladerunner asks whether an artificial being can be distinguished from a human, and what that distinction means.
A similar question appeared in the movie: Her (2013). Her explores an authentic emotional experience arising from a relationship in which one participant is artificial (not Commodus).
Later, Westworld (2016) repeatedly destabilizes the distinction between programmed behavior, memory and lived experience. More synthetic activity creates greater uncertainty. Greater uncertainty encourages more automated verification. More verification creates “friction” for legitimate participants. And because automated content is cheap, the volume can keep rising, even as the value of individual interactions falls. In Westworld, it fell pretty hard.
If we look at hiring practices, a common theme is AI writes resumes, introductory letters and PhD theses. Then an AI on the corporate side rejects the nicely parsed resume as AI, based on a hidden evaluation of proofs. HR says: “We” can’t find good people. There is no human in the loop. It’s funny, until the AI starts ghost writing the “yearly goals” section of your inactive digital twin (Severance 2022). AI collects conversations from pseudonymous Reddit users to validate questions, but those conversations are already infiltrated by incredibly plausible AI bots, with far higher resource allocation than most users. Our team recently rebuilt the company website using AI tools. But, does an automated process direct “real” people away from a “synthetic” website, because of hyphenated run-on sentences?
Activity is no longer a reliable proxy for attention
“Activity is no longer a reliable proxy for attention. Attention is no longer a reliable proxy for commercial value.” Each is a proxy for the next, not proof of it. When the system starts optimizing for the proxy rather than the thing it is supposed to measure, it can report improving performance while the underlying value deteriorates. “We got 150 clicks on our most recent post.” Ok, did anyone purchase (the verification layer)?
IoT exists in physical reality
A little book, Simulacra and Simulation (1981), was written many years ago by Jean Baudrillard. The book was a featured prop in the Matrix series (1999). Jean examines the relationship between representations and a grounded (intersubjective verification) reality. IoT exists in physical reality, but software remains in cyberspace. What happens if my AI decides, WarGames (1983) style, to input numbers that create a cascade of real-time, real-world results? The trail of a nefarious actor could be hidden with misdirection, just as soon as it appears, Keyser Söze style (The Usual Suspects, 1995).
“Outside, London goes about its business. Carriages pass. People make appointments, keep promises, tell lies, buy things, fall in love and occasionally write an absolutely dreadful letter to the editor. And somewhere, a machine is attempting to determine whether any of them actually exists.”
ChatGPT
We’ve touched on trust, authenticity, accountability, economic value and the possibility that a system can become extraordinarily efficient at producing an appearance of the thing it is meant to deliver.
We had this concept a number of years ago, when marketing became a full send “automation exercise.” Vast amounts of corporate cash spent on expensive enterprise level “full solution MAR-COM tools” that built out plausible sequences of interactions. You were told to personalize, filling blanks with little snippets of relevant company info to trigger a human connection, while logging every click on the route to a sale. You can supersize, go for hyper personalization, powered by AI> The PA never bought a thing.
“Implementing hyper-personalization requires a robust data infrastructure and a commitment to data privacy.”
IBM
Personalization is now the thing to avoid, because you might trigger a human reaction. That’s if the AI intermediary is not triggered first by your weak punctuation. In War Games, “WOPR” (1983) eventually discovers that the game has no winning strategy. Our contemporary systems may be operating in environments where the participants are rewarded for continuing to play, regardless of whether the game still serves its original purpose. “A strange game. The only winning move is not to play,” Joshua.
