The AI apocalypse is postponed. It will be rescheduled after someone can figure out which fuse has blown and the automatons have learned to reboot themselves.
The AI Apocalypse Primer
It seems as if we are once again in the grips of a panic over artificial intelligence and an end to meaningful work. I started thinking about this a couple weeks back when a friend shared his thoughts on FaceBook regarding the downsizing of the workforce on account of automation and Artificial Intelligence. I was, politely shouted down as being out of touch on the topic. So, I’m taking my marbles home. (I know. No one plays marbles anymore.) I thought this might be a better place to share a few cogent thoughts on the topic.
This is hardly a new topic for me. I have been reading about real AI for decades. Even then, many of my references are even older. So, it’s safe to say that I’m not newly introduced to the topic. You can find a great discussion in Douglas Hoffstadter’s Metamagical Themas. He predicted OCR advancements and dynamic font fitting (similar to early Adobe Acrobat technology) decades before it happened. Going a step further, he explores ideas related to artificial intelligence and artificial life. If you like this kind of thing and haven’t read it, you should. More recently (circa the first Internet Bubble) is a book by an economist, Jeremy Rifkin called The End of Work. I don’t want to spoil the plot, but the title gives away his main opoint. The culprit is technology and automation. Just for good measure, we can toss in a dose of Thomas Friedman’s book, The Word is Flat. Those texts should serve as a primer on this topic and from three separate but past eras.
Depictions and Predictions of the AI Apocalypse
The problem I have getting worked up about AI is that it’s not news. Mary Shelley created the original AI boogie man in her novel Frankenstein. Clarke had us all in fear an automated doorman when HAL9000 refused entry to David Bowman back on his ship, Discovery.
I believe the scares that came to us in the dark of the movie theater had the most powerful impact. The silent era seized the anti tech position in the epic film Metropolis. It’s a sweeping vision of a future devaluing the humans who supply labor. It forged the template for technophobia which is followed still.
Trust the Company and Sanctuary
Talkies brought more paranoid nightmares of robotic rampages and sinister computer systems in Forbidden Planet, 2001: A Space Odyssey, Logan’s Run, West World, The Stepford Wives, Joshua in War Games, Ashe and Bishop from Alien, the replicants from Blade Runner, The Terminator, the film version of I Robot and everyone’s favorite distopia, The Matrix.
The androids and self aware mainframes in these tales lie and cheat. They become psychotic or at least homicidal for a cause, even if it’s only a game. They treat us like sheep or worse, but all the characters were constructed intentionally to scare us.
In sharp contrast to the fear pedaling we also have positive images which have not assuaged the public distrust of AI. From The Jetson’s Rosie, and Number 5 from Short Circuit we get a few laughs. Star Trek‘s Data, Star Wars Droids and Spielberg’s AI become metaphors for minority rights. Most interesting to me though, are Baymax from Big Hero 6, Robot from Robot and Frank and Samantha in Her.
The New AI Model
The creators of Baymax, Robot and Samantha wrote and directed technology based tales as tech-literate people for a tech-literate audience. It’s a significant difference that sets a new expectation from technology. The characters don’t act like they were programmed by homicidal lunatics. They don’t intend to cause harm. Most importantly they don’t intend anything on their own. Their heroism, sacrifice and warmth are artifacts of humans projecting their hopes on them. Their personalities are engineered. They can all be turned off with a switch. They are tools who are perceived, because of our human perceptions, as having traits of genuine emotion and intelligence. The message evolves into we control technology, not the other way around.
Still fear of automation and AI persists. The latest wave of paranoia comes from professionals feeling global market pressure where none previously existed. It reminds me of what I saw in the 1980s and 1990s as the printing and commercial art industries expelled talented and experienced craftspeople in favor of kids with keyboards. Fortunately, I was a kid with a keyboard.
Today, automated systems are taking dictation, answering phones, examining documents and looking over the shoulders of people in professions who relied on staffers. Even medical opinions are networked globally over a simple x-ray. Like typesetters, film strippers and many others before them, doctors, lawyers, and other white collar workers feel the encroachment of automated productivity.
A new wave of prognosticating doom is fueled in the media by celebrity scientists and technologists predicting the worst possible outcomes. It’s important to understand that every time a celebrity expert gets the run of a relatively slow news cycle, they are almost always pedaling a new book, movie or lecture tour. No one cares very much if you’re a pundit who likes our current heading into the future.
The proof is in the products.
Whatever we think of our digital AI overlords, they simply don’t exist. In reality, some of the technology has promise to aid human endeavors. Like previous waves of industrial and digital technology, there will be massive reassignment of roles, but we won’t be displacing the entire workforce at once. As the automat has shown, we still want and need to interact with other people. Extraordinary claims require extraordinary proof, which doesn’t yet exist. How can I support my optimism in the shadow of the AI apocalypse?
What inspired this missive was a claim that most mundane and specialized tasks will be performed by intelligent machines, instead of humans. Not only will labor intensive work be automated, but even professionals will see demand cut significantly by emerging technologies. The only imaginable consequence, according to the proponents of this school of thought, will be insufficient demand for human labor and increasing waves of unemployment.
Outside of a few celebrity rants, the evidence for this claim remains thin. I suppose the Goliath to confront first is the fabled Watson computer system that decimated the best human players on national television. Nova produced an entertaining documentary on the development of Watson. Watson as seen on television is just a really good search engine. Like a savant with an almanac and a set of encyclopedias, it can answer trivia questions. I really hope that it serves professionals in medical science (one of its promoted applications) with a valuable resource for mining useful information from the internet. On a scale of real intelligence though, it doesn’t hold a candle to Dustin Hoffman’s Rain Man or even a dog. It will accept any input and weigh it based on algorithms against other input, likey flat earthers and other oft repeated and unsubstantiated claims. It has no other way of judging the value of information and has produced some amazing boneheaded answers as well as obscenity laden rants. Microsoft’s experiment with a chatbot called “Tay” took only a day to start parroting racist and bigoted output. The only way that either system knows that the underpants come from K-mart is by scraping the IMDB.
Self driving cars have been in development for a while. Google has made great progress and interesting vehicles, but there’s a long way to go before they become a norm. The first self driving car collision due to computer error has only recently been reported. If the public reacts to these vehicles like they did with Microsoft’s Tay, expect insurance claims to skyrocket and a cottage industry of playing chicken with rolling computers to be the latest get rich quick scam. I think that professional drivers with trucks, cabs or even Uber are safe for a while. There are some promising applications of this technology. It came from DARPA and I think driving in war zones is a great way to use a self driving vehicle. I’m not sure I trust the technology in Miami or Atlanta during afternoon rush hour.
In spite of the awesome PR from MIT’s media lab, the most successful AI that most people are likely to encounter are telephone transcription systems, smart phones and the Roomba. Sure, a cat riding a Roomba makes for a really funny video (Google it yourself), but if you move the charging station, your little sweeper starves to death within reach of a recharge. Ever get lost in phone jail because someone spoke too loudly in the background? As for the smartphones, I know people who can not get their phone to understand much of anything they say. I consider myself lucky in that regard, but some of the auto correct errors and misunderstandings I have seen on my own screen, though hilarious, are unfit for this venue.
When I was in college, I spoke a strange language. I knew about computers but if I spoke of hardware, most listeners thought of hammers and wrenches. A web came from deceit or spiders, and spiders were nouns, arachnids, and not computer programs or what they did. Not only do I believe in the resiliency of humanity, but I believe with some certainty that any plan for Skynet or the AI apocalypse by any other name is ridiculous. TV did not kill the movies or radio. Desktop publishing didn’t kill graphic design or printing. In fact, paper and ink sales continue to break records. The internet didn’t kill the library or books.
Today, we have designers for print and web. We have content creators and consume more photography than I ever imagined. There are social media experts, SEO, bloggers, CRM, reputation managers and the list goes on and on. The only constant is change. Everyone harbors a little fear and trepidation over the uncertainty of the future. On the other hand, I look at the bounty created by technologies for the past forty years. For over a century, productivity leaps have threatened to overthrow the need for human workers. The science of electricity created tales of monsters. In reality the only outcome has been more productivity than we ever imagined. So, the AI apocalypse is postponed to an indefinite state of “TBA.”