Mind you, this was a generative design algorithm meant to assist pilots to land a plane. But for this AI, it just became a game.
NOTE: the following are snippets from a (very) rough cut draft of my book-in-progress. It is a long-read. I am often accused of employing “TL;DR” blog posts – an army of phrases, moving across a wide landscape, to wrestle my ideas to the ground. So be it. Maybe a book would be better to get it all out there.
22 August 2019 (Serifos, Greece) – “Technology blinds us with fairy dust. Technology will, metaphorically speaking, change at midnight, leaving us dressed in rags and missing a shoe. Technology is being built by people who are much more intelligent than me, but none have my best interests at heart.” – Peter Stannack, CEO, Prebbler Technologies (and self-described “Punk AI Proponent, Machine Intelligence Integrator” but I’d also add polymath)
Introduction
Many of us tech geeks are romantics – we spend a lifetime looking for technology that will solve our problems. We “believe” in it. But I suspect everyone is a romantic at heart however much they try to keep it hidden:
Parent of golden dreams, Romance!
Auspicious queen of childish joys,
Who lead’st along, in airy dance,
Thy votive train of girls and boys.
– Lord Byron, “To Romance”
Lately it seems all the tech world can do is create a new herd of incurable problems that are worse than the problems meant to be “solved”. As Peter Stannack and I have discussed many times in a string of email chats, the irony is that the promise of technology as a guarantee of future well-being has become increasingly powerful even as the future itself is increasingly insecure.
I suppose at some point in my life (years and years and years ago, in my Palaeozoic period), I fervently believed in “happily ever after”, that technology … being built by people much more intelligent than me … always had my best interests at heart. The romantic in me had spent a long time looking for technology to solve our problems. Not all, but many. And I saw the relation between technology and romanticism. There was this patina of something being pragmatic and optimistic. But our essential problems are within us, not in our relations with the external world. Technology will never fix those.
And so, as this new herd of incurable problems grew worse, all the romance left.
But more ominously, technology has released a monster from its cave, and that monster shall not return to its cave. The weaponization of data, of information is mutating at alarming speed. Communication has been weaponized, used to provoke, mislead and influence the public in numerous insidious ways. Disinformation was just the first stage of an evolving trend of using information to subvert democracy, confuse rival states, define the narrative and control public opinion. Using the large, unregulated, open environments that tech companies once promised would “empower” ordinary people, disinformation has spread rapidly across the globe. The power that tech companies offered us has become a priceless tool in propagandists’ hands, who were right in thinking that a confused, rapidly globalizing world is more vulnerable to the malleable beast of disinformation than straightforward propaganda. Whatever we do, however many fact-checking initiatives we undertake, disinformation shows no sign of abating. It just mutates. And it keeps perfecting itself. More in my concluding thoughts at the end of this post.
a single data point, is not captured and stored. Here, a baby has its own camera, capturing memories.
Those of us of a certain age (I am 68) were fortunate to be at the dawn of digital history, the birth of the newly created digital computer. It was called the “electronic brain”. And, we were told, “it can do … EVERYTHING!”
And the first person to write a FORTRAN program for that new area would pick up the gauntlet and claim “just amazing results compared to what came before!” And so, with careless extrapolation, every engineering or mathematical problem in the world seemingly would fall to FORTRAN speaking electronic brains.
But with time (and lots and lots and lots of experience), the limitations became very apparent. Further progress was expected but not exceptional. Today, AI and machine learning have a similar smell to it. But we are told “HOLD ON! It really is different this time! We have all of this … DATA!!”
The other day Peter reminded me of a scene in the Woody Allen movie Annie Hall. In it, Allen tells a joke:
Two elderly women are at a Catskill Mountain resort, and one of them says, “Boy, the food at this place is really terrible.” The other one says, “Yeah, I know, and such small portions, too.”
Peter’s point was simple. This seems to be how data scientists feel, too. They want much more data. Even if it is bad data. Because data will (eventually) solve all problems.
Consider the rapidity and scale of this digital data inversion – its “hockey stick” profluence – and then ponder its implications for how we understand and live in the world. Data now floods our planet oceanically, with consequences – for data science, data storage, data networks, processing technologies, artificial intelligence, social media, smartphones, robotics, automation, the Internet of things, autonomous vehicles and weapons, and the surveillance state – that no one could have imagined two decades ago.
Nearly every profession and discipline has become data-dependent, if not data-overwhelmed, each now vying to become its own avatar for a “thin-slicing” paradigm (itself inverting Malcolm Gladwell’s conventional “blink” conception of thin-slicing) that sucks unfathomable quantities of layered information about our material world (pretty much anything that can be measured, with much of it also imaged) into its maw.
We can profitably compare the psychological impacts of the data revolution of the early 21st century to the interior emotional and cognitive consequences of the printing revolution that originated in the early 15th century. During this era of “mechanical reproduction” (about which Walter Benjamin had famously written), book production increased from about 20 million volumes in the 15th century to more than 1 billion by the 18th century, with unfiltered access to information creating new dramas of personal subjectivity associated with epistemic confusion and social anarchy. Sound familiar? Now compress that growth from 400 years to 20 years.
Or compare the disruptive economic and labor force consequences of the data revolution of the early 21st century to the cataclysmic consequences for labor of the steam-powered industrial revolution of the early 19th century. EP Thompson’s Making of the English Working Class reminds us of the political layers to these transformations, with innumerable fascinating analogies, such as the Benthamite panopticon (mirroring the modern surveillance state) and development of an “outworker” industry within the textiles trade that deskilled and immiserated workers, (resembling the modern “gig economy”).
But there are two big problems here, and I want to quote Peter again (he has thought through these issues for years, and we are on the same page on all of this):
1. The first is that quantitative measurement can be seen as treating facts as unproblematically existing in the world. This means that data scientists are accused of of treating people like objects … and failing to capture and interpret the many complex properties which inform human activity and constitute social life. And so what you have is bad propagation because these quantitative measurements do not capture the human complexity and distort the reality. It’s just like anomalous propagation or “false echos” in radio when the transmitted signals are distorted because the system did not factor in environmental conditions.
2. The second is that data scientists do not (normally) recognize that so-called “facts” are not unproblematic. They are the outcome of implicit theories about social life and social relationships —theories that are obscured by claims to neutrality and objectivity.
This is not to say all this data analysis has not been a benefit in many regards. Just one example: before I went on holiday I managed to squeeze in the Google Analytics summer party in Paris – one part “hey, stuff we are doing!”, one part “have a drink!” What impressed me the most was Google Maps, a platform that has both spawned and leveraged amazing geospatial mapping innovations based on the sharing and exchange of location-specific data – built structures, geographic and geologic features, demographic attributes, moments in time.
Google Maps support for data layers has opened new opportunities to integrate different types of data by overlaying them on the same location grid, creating possibilities, not simply to virtualize three-dimensional space, but to attach endless numbers of other data “dimensions” to this space, creating a vertical landscape that in some instances approximates holography. The Google Maps data layers create an n-dimensional, data-rich mapping “landscape,” which includes location-tagged traffic and transit data, photos, weather information, and Wikipedia articles. More than eight million websites now use the Google Maps APIs, making it the most ubiquitous Web API globally.
Oh, and here is your “Weird Fact of the Day” which I learned at the Google fest: cloud computing leader Amazon has for years used super-secure trucks to transport hard drives with business data to its cloud computing data centers: actually a forward-looking use of an “on-its-face” inadequate (if not immediately anachronistic) delivery mechanism to secure the company’s first-mover advantage.
Or astronomy, which is my second study passion after the sea. For instance, machine learning now exceeds the ability of astrophysicists to classify space data, such as space spectral photography, to the point that deep learning models generate images that are amazingly realistic to the point of being indistinguishable from real photos.
Or look at XML, and its successor markup and tagging languages, which have made possible new ways to “liberate” content (describe, extract, separate – semantic and otherwise) from its analog and digital forms, allowing for limitless exchange, recombination, reuse, and reconceptualization of this content, a kind of virtual Big Bang. It transformed the legal technology industry. It provided that industry with advanced statistical methods and machine logic to review and analyze unfathomably large and generally unstructured, distributed data repositories. It provided legal technology with the means to use natural language processing. Granted, legal technology is downright primitive when viewed in the general sweep of technology. But it will improve.
Note to my e-discovery readers: you have Robert Mercer, the American computer scientist and early developer of artificial intelligence, to thank for all of this. Yes, the same guy who funded Donald Trump, the guy who funded the start of Cambridge Analytics, etc., etc., etc. It was he and his partners who (beginning in 1972 at IBM) invented the statistical machine translation software that would be the basis of not all machine translation software to come. And as that technology and those techniques propagated through tech, would lay the basis for all advanced statistical methods and machine logic to review text – and give birth to computer assisted review/technology assisted review. But as Hyman Roth might say:
Oh, and Mercer and his crew made millions of dollars in the process. Well, they made billions, really, because the team would go on to use those millions to fund the most successful hedge fund in the world.
Yes, other human-readable and machine-readable languages for structuring and exchanging data have since largely displaced XML, especially in cloud-sourced Big Data applications. However, the early adoption and influence of XML helps us to confront new dimensions of the complex historical and philosophical relationship between content and form, between the substance of things and the shapes they take.
And it ushered in the virtual world that now both mimes and shapes our reality. Its impact also represents a paradigmatic instance of the process by which technology revolutions can bend the arc of history. The Big Data revolution is the output of a technology innovation equivalent to – and sharing elements of – other innovations implicated in pivotal moments of human history.
And therein lies the problem. This is more than just a technological innovative moment. These are Archimedean moments, creating entirely new tools kits and entirely transforming the perspectives on ourselves (as subjects) and on the world (experienced as the set of objects with which we interact). In this sense, these revolutions are innovations far, far beyond the use of material. We are flipping on their heads all previous notions of the cognitive and phenomenological. This is the equivalent of the discovery of fire.
And it differs in one other critical respect from prior innovations. The emerging landscape of machine learning and artificial intelligence applications requires inputs of non-material, invisible, unfathomably large quantities of electronic information harvested from business, social, and sensor data universes. The data explosion is truly turning both our emotional lives and physical bodies inside-out, with surveillance, tracking, and asymmetrical information landscapes fundamentally reshaping our identities and capacities.
A brief quote from my video interview with Peter Schwartz who was the Senior Vice President of Strategic Planning for Salesforce.com and responsible for doing that company’s “big thinking”:
Data-driven ride-sharing services such as Uber and Lyft preview what our new world is going to look like, with new realities determined by information asymmetries and one-way information interactions that allow these companies, via feedback loops, to “game” and condition behaviour – in this case, driver behavior. The enticing language of choice, freedom, and autonomy used by ride-share companies to recruit drivers masks the extent to which these companies algorithmically program and constrain driver choices. It is spreading all over.
We locate mystery, fear, confusion – other words for God – in the unfathomable. The micronic scale of subatomic particles and the eternally vast spaces of the universe are beyond our capacity to process. Similarly, data volumes – terabytes, petabytes, exabytes, zettabytes and beyond – have scaled beyond our capacity to process as individuals.
And we are freaking out at this “pivot”, from the ability to understand what it means to be human, to not understanding what this new Data God is doing to our individuality, agency, autonomy, judgment, memory, and desire.
It’s the generational piece of this story that most interests me, with children, grandchildren and now a great-granddaughter. What percentage of one’s life occurs before and after the temporal “digital divide?” How does this balance between a personal foundation – the embodied self – hatched before and after the advent of the web browser and the smart phone matter? Will any concept of embodied selfhood as we currently and historically understand it, even exist? What will personalities and emotional states of children born into this emergent digital grid look like?
But we are told “Do not worry. Data is the only thing that makes us safe, makes us understand”. And yet it puts us in danger. Because as Peter Stannack observed:
We must have data about “you”, for instance, to improve your life, make you better. Oh, and also to make sure you are not going to do anything naughty. And we have to feed “you” with an incredible data diet – no, I mean an “incredible” data diet in the sense no one could believe it – to make sure you “understand” things, because otherwise you cannot. Oh, and as an antidote (snicker, snicker) to fake news, deepfakes and weaponized information.
But don’t worry about that data dependency … and the tendency of errors to propagate. It’s special. “We’ll fix that. It’s data science. You can trust it.”
But let’s move on to some of that propagation.
AI and the fabric of computing
The forward-looking science of complexity has quickly grasped the historical significance of this Big Data revolution -Big Data, artificial intelligence, and machine learning – and how it is creating a great transformation, upending our epistemological fundament. They have become a force multiplier.
And we are learning that AI AI doesn’t “think”, really. It evolves.
In the 1950s, the psychologist J. P. Guilford divided creative thought into two categories: convergent thinking and divergent thinking
1. Convergent thinking, which Guilford defined as the ability to answer questions correctly, is predominantly a display of memory and logic.
2. Divergent thinking, the ability to generate many potential answers from a single problem or question, shows a flair for curiosity, an ability to think “outside the box.” Guilford used this as one example: it’s the difference between remembering the capital of Austria and figuring how to start a thriving business in Vienna without knowing a lick of German.
I do not have space to properly describe Guilford’s work … fully expressed in his monumental and highly influential text The Structure of Intellect Theory … but suffice it to say his theory comprises up to 180 different intellectual abilities organized along three dimensions: operations, content, and products.
When most people think of AI’s relative strengths over humans, they think of its convergent intelligence. With superior memory capacity and processing power, computers outperform people at rules-based games, complex calculations, and data storage: chess, advanced math, and Jeopardy. What computers lack, some might say, is any form of imagination, or rule-breaking curiosity—that is, divergence.
But as Derek Thompson notes in a series of articles that have run for the last two years in The Atlantic magazine, we have it wrong:
But as Derek Thompson noted in a series of articles that have run for the last two years in The Atlantic magazine:
AI’s real comparative advantage over humans is precisely its divergent intelligence—its creative potential. One of the more interesting applications of AI today is a field called generative design, where a machine is fed oodles of data and asked to come up with hundreds or thousands of designs that meet specific criteria. It is, essentially, an exercise in divergent potential.
Generative design. Which opened this post. The algorithm that was supposed to figure out how to land a virtual airplane with minimal force as part of an automatic pilot program. But discovered that if it crashed the plane, the program would register a force so large that it would overwhelm its own memory and count it as a perfect score. Presumably killing all the virtual people on board.
Oops.
Yes, there are many benign examples. One of my favorites is a simulated robot that was programmed to travel forward as quickly as possible. But instead of building legs and walking, it built itself into a tall tower and fell forward. How is growing tall and falling on your face anything like walking? Well, both cover a horizontal distance pretty quickly. And the AI took its task very, very literally. Why walk, when you can just fall? A relatable sentiment.
I know what you are thinking. Just more evidence of the dim-wittedness of artificial intelligence. No. Wrong. Actually evidence of the opposite: a divergent intelligence that mimics biology. Taken in broad strokes, evidence that evolution, whether biological or computational, is inherently creative and should routinely be expected to surprise, delight, and even outwit us. Because evolution provides a creative fount of complex and subtle adaptations that often surprise the scientists who discover them.
However, the creativity of evolution is not limited to the natural world: artificial organisms evolving in computational environments have also elicited surprise and wonder from the researchers studying them. The process of evolution is an algorithmic process that transcends the substrate in which it occurs. Indeed, as noted by Dusan Misevic of the Center for Research and Interdisciplinarity in Paris:
Many researchers in the field of digital evolution can provide examples of how their evolving algorithms and organisms have creatively subverted their expectations or intentions, exposed unrecognized bugs in their code, produced unexpectedly adaptations, or engaged in behaviors and outcomes uncannily convergent with ones found in nature. Such stories routinely reveal surprise and creativity by evolution in these digital worlds, but they rarely fit into the standard scientific narrative.
Instead they are often treated as mere obstacles to be overcome, rather than results that warrant study in their own right. Bugs are fixed, experiments are refocused, and one-off surprises are collapsed into a single data point. The stories themselves are traded among researchers through oral tradition, but that mode of information transmission is inefficient and prone to error and outright loss. Moreover, the fact that these stories tend to be shared only among practitioners means that many natural scientists do not realize how interesting and lifelike digital organisms are.
But you see the issue.
Yes, evolution provides countless examples of creative, surprising, and amazingly complex solutions to life’s challenges. Some flowers act as acoustic beacons to attract echo-locating bats, extremophile microbes repair their DNA to thrive in presence of extreme radiation, and bombardier beetles repel predators with explosive chemical reactions. Many more examples abound, covering the full range of biological systems.
However …
CONCLUDING THOUGHTS
The advent of Big Data, artificial intelligence, and machine learning have rapidly encroached upon (inhabited, invaded, colonized, body-snatched, pick your verb) the biological and social constituents of individual human identity – the language, reason, judgment, and desire of the embodied, conscious self. Data collection and extraction has in short order hollowed out and inverted – metaphysically if not literally – the physical forms and landscapes through which we have always, as a species, mapped our reality. Within several more decades, it now seems clear that digital data and artificial intelligence will reconstitute and reformat what it means to be human. It belongs in this conversation.
[insert]