Blog: Fahrenheit 2050: the Future Shock of AI & Machine Learning
Can otherwise benevolent AI hearken our ultimate undoing?
I’m categorically ambivalent on the subject of technological advances/intrusion into our everyday lives. That includes AI and higher machine learning. On the one hand, AI is used and will be used for brilliant life-changing developments in endeavors like medicine, generative design, scientific research, and a lot of what is to come of which we really have no idea.
Yet, mostly it will be used for harmless mundane things like selecting which clothes to wear or what music you want to listen to — a tool of software development. Anything developers can make a buck with. AI already is deployed everywhere on the Internet in apps developed by Amazon, Spotify, Google, Apple, and countless others, second guessing our buying habits, and convincing us to adapt to subliminal or blatant suggestions.
“However, it’s the flip-side of AI — it’s potentially sour underbelly that should give us pause if it hasn’t already creeped us out a bit.
As AI engines become more powerful they are designed and put to use for more sophisticated (and redeeming) purposes, such as conducting a mobile combustible engine along a freeway or managing a devastatingly complex medical procedure. A case in point are IBM’s numerous and brilliant Watson platforms.
The passive and tacit acceptance of AI into our daily lives doesn’t mean that AI is not insidious behind the scenes. We are wary of the early days of the Clinton Whitehouse Internet surveillance programs — such as the infamous Clipper chip. He thought it was a jolly-good idea for the defense and law industries who — as it turned out — were the only avid celebrants of the technology. Clinton wanted to impose the encryption which could (supposedly) only be cracked by hackers and the CIA. The chip itself was quickly and unceremoniously hacked upon it’s beta-release to the NSA.
“The backbone of the commodity of AI is not unlike ‘90’s’Clinton spook programs.
The monetization of AI resides in the IoT and SIMM cards of our digital-existence. AI engines power device applications for just about every mundane quotidian one can imagine. Toys, home lighting and climate control, refrigerators, even pets and personal companions, all of whom are perfect for hosting backdoor video and audio surveillance, and worse, the hacking and abuse of such information.
“Most people are painfully ignorant of such trespasses of their privacy making perfect targets and paying for the privilege.
AI and machine learning are often conflated with human intelligence. A machine can’t ‘think’ in the way we can. It’s not going to invent anything we haven’t taught it or derived from a data-set we have fed it. It’s not any smarter than us — only it can ascertain facts beyond our capacity or patience.
It can develop algorithms that it uses to predict our wants and needs when we’re too lazy to make the effort. Within its compass it can do an awful lot of what we want it to do. At some point, AI engines will become our proxies and avatars, depersonalizing us to ourselves and to others. We can expect nothing redeeming from such slothfulness.
The temptation to overstep AI boundaries is already a divisive issue. Jobs and occupations that were erstwhile humbly managed by humans have already been usurped by AI and robots, such as fast-food and customer service, mundande vocations both of which AI can perform standing on its head. But then there always seems to be some overreaching Elon Musk-oid industrialist who sees yet more (monetized) purpose for AI.
“I relish the eventuality my phone will someday bellyache “insufficient data.
These concepts invariably cross ethical lines of what we used to call “humanism.” the notion that occupations and pursuits that require sentience and presence has a habit of making one’s skin crawl. For example, AI psychoanalysts, lawyers, suicide hotlines, judges, priests — all of these unequivocally cross the line.
Musicians, artists, dancers, actors, screenwriter, authors all can be dispensed with, as can the outgo to finance their works. Some of that is achievable rather easily given the simplistic and reliably predictable output of human nature. A musician can begin with a few bars and Piano Genie can guess the rest. Some of the priciest artworks hitting the market are entirely AI created– submarining those painter and sculptors merely trying to scratch out a meager living.
“Not to be taken in by gleefully smug marketing programs, AI developers aren’t any more ethical than the cheapest scumware.
Case in point this post’s frontispiece Portrait of Edmond Belamy, 2018, created by GAN (Generative Adversarial Network) — (yes “Adversarial” Network”) sold for over $432,000. That shocking sum -45X higher than the piece opened at auction could easily float an artist’s family and vocation for years. I bet you dollars-to-donuts it was derived from Francis Bacon images: AI simply cannot make something out of nothing. In this way it can be said that any AI generated art is in essence a knock-off of its human betters.
When money is the impetus, little will our trepidation matter. The AI juggernaut is poised to take over just about every aspect of our existence. Because too many of us empower our devices with massive personal big-data we run the risk of egregious oversharing. For example, we already allow AI to use our personal data for predictive purposes or to guide our behavior and decisions. Many of these are by no means trivial.
A total embrace of AI in our lives would require full disclosure and permission to make itself far more insidious than we can imagine. Using data from a mere handful of apps, AI can learn when we are likely to sleep, eat, bathe, have sex, walk, talk, come and go, as well as a full spectrum of bodily functions and biometrics.
“Siri, which bridge has the best view for jumping off?
All of which constitutes an overinvestment and codependency that bodes ill for all concerned. The more we overinvest in AI, the less thinking and acting we do for ourselves. The more dependent we become on AI, the more shitless and pathetic we become when there’s no Internet connection.
I have written lately how AI is the future of virtual (online) dating. Owing to the cumbersome requisite investment of time waste virtually-dating, it will be an inevitable and painless jump. What might that look like?
Every effort that you and the dating platforms make is just a starting point for AI engines or pimp-bots. From there, AI expands from one-dimensional (data), to more intuitive (recognition) machine learning processes. AI driven apps would automatically use your member profile criteria to search for people whose profile contains keywords that meet that criteria all on its own, freeing up valuable hours of free time to (gasp!) go out and allow an IRL encounter to come of its own accord, if it should come at all.
Consider this hypothetical scenario in the not to distant future: an AI-driven virtual dating app we’ll call the AI engine ‘Kiki.’ I’m very persnickety, so I want to be very specific in my terms of engagement:
“Kiki, show me single women within 25 miles, between 48 and 52 years of age, atheist, thin (under 120# at 5’ 6”), bachelor’s degree minimum, divorce or widowed, with no children at home, who isn’t a boozer, and makes at least $25K yearly salary.
“She mentions at least one of my favorite authors, musical artists, and likes museums, travel, hiking, dining, and doesn’t smoke.”
So far, the platforms can keep up, but that’s as good as it gets. From this point, is where — with a bit of machine learning — AI could become very specific, and troll in a much more sophisticated way. From parameters set by the user, AI can recognize and analyze certain facial expressions, postures, and other subtle nuances that are more humanistic than the vital stats in your profile.
“Most developers regard privacy and security scripts in their code as counterintuitive to their pocketbook.
She does this in tandem with an enormous database of research on every known facial expression, posture, or gesture that has ever been digitally recorded. Future dating profiles will necessarily need to include audio-video footage, which in terms of presentation would be leaps and bounds above the status quo. You will note, that audio-video profiles have been around, but haven’t really ever got any traction. Again, poor execution
When people have audio-video media in their profile, Kiki will be able to evaluate fine and gross motor function: “must be graceful and move through space like a ballerina — mannerisms, facial transitions and eye movements, vocal tone and elocution, “the way she turns her head like a lithe French mademoiselle. She also can make more accurate depth perception cognitions and fewer assumptions of the structure of a body in three-dimensions.
Match, OK Cupid, POF — obsolete, and binnablemedium.com
In this way AI could navigate thousands of prospective dates for you. AI can easily be programmed to mimic the tiny lexicon and dialog that colors most any early virtual-dating transaction. Men could send dick pics to every blonde woman on a given site without even knowing it. It can also carry on quite sophisticated dialog virtually undetectable to the average bear. (FTR, for Americans, this is a 7th grader’s reading prowess.) If a proposition should be dignified with a response, AI can alert the user who can then deign to carry on in person, as it were.
“AI can order milk when it is run low, or order condoms when we need them: but is that what we really want?
Some would concede their most sacrosanct personal data willingly to marketers and vendors in the false belief that their lives will measurably benefit from such intrusions. Nothing wrong with that — a chacun son gout. To a better end, AI also helps those struggling with disability to navigate, assist, monitor, medicate, or otherwise be deployed to empower those who really need it. That’s AI put to good use.
However; IoT makes possible myriad opportunity and avenues of surveillance and data-theft that far exceed our wildest imagining. Any device can be hacked: a hacker can unlock your home, turn off your furnace and let your pipes burst, shanghai your self-driving car, post malicious social media about you and copy everyone on your phone, and in some cases hack medical devices that keep people alive.
But it’s not twenty-first century human nature to only develop ethical software. At the same time, shaming, identity theft, and blackmail rates will skyrocket as developers churn out applications with little or no firewalls. The notion of establishing a database of our every breathing moment blatantly there for the taking of any hacker who can crack thee 1-2-3 default password so cleverly built-in to our devices, or rather surveillance conduits must be taken as a serious concern and misgiving of mainstream AI.
For myself, I’m happy to leave most of my incidental behavior and experiences to chance, managing on my own the balance of cognitive performance. I don’t need a proxy to think, predict, second-guess, suggest, proposition, arrange banking, prognosticate, or otherwise jerry-rig my quotidian, thank you. Although I can’t crunch my own biometrics at trillions of instructions per second, I am full well capable of fucking up my life all by myself.
At least there’ll be no spoiler endings and an element of surprise, hope, and open mindedness that AI deceptively narrows even as it purports to broaden our perspectives and outcomes.