The Power, Potential, and Perils of AI

New York Before Cars

In the late 1890’s, New York had on its streets somewhere between 150,000 and 200,000 horses.

Horses were a reliable, steady and proven technology that carried the city’s commerce and capital in an animal driven stream of reliability and dependability.

But horses also brought with them a significant downside. Each horse, as a by-product of its existence, produced 30 pounds of manure and a quart of urine each and every day. Most, if not all of it, wound up on the city’s streets.

By the start of the 1900’s, vacant lots all around New York featured manure piles 60 feet tall and higher, and with each of these festering lots came billions upon billions of disease infected flies hatching and spreading illness throughout the boroughs.

In the summer, the dried and airborne dust from the ground-up manure on every single roadway damaged eyes and irritated respiratory systems, and in the fall and winter, the wet runoff from the streets would seep into basements and bring with it all the hardships that one can imagine might come from a cellar full of liquid defecation. 

On top of this, over 15,000 horses died each year, often simply from overwork, and more often than not they were simply abandoned in the city streets right where they expired, left to rot, decompose, decay and stink up the area unchecked for days and weeks until the city reluctantly would come by and remove their corpses.

This problem though was not just limited to New York, every other major or growing city in the world around this time was experiencing essentially the exact same thing.

Where Have All the Horses Gone?

Then a couple of innovations occurred, first in the form of a new technology called the electric streetcar, and then, in 1907, by way of a new type of moving assembly line that allowed for a novel form of affordable mass produced product to be widely adopted – the automobile.

By 1912, cars outnumbered horses on the streets of New York. By 1917 the last horse-car was put out of service, and by 1922, Assistant Fire Chief Joseph “Smokey Joe” Martin rang the alarm bell at the station for Fire Engine 205’s final outing for when a whip would crack and reins would be snapped as Buck, Bellgriffin, Eamybeg, Penrod and Waterboy, the loyal horses that for years and years had pulled the fire brigade’s engine all over the city, were officially retired and replaced by an iron steam engine and hose wagon.

This was 100 years ago.

There’s a good chance that if you have grandparents who are alive and that you make fun of for still using flip-phones, they probably have memories of the those last days when horses were phased out as a significant transportation technology, as well as of the dire warnings of impending chaos that were predicted to occur as a result of the loss of this once dependable means of conveyance by the futurists of the day.

Looking at it with hindsight, who knows, they may have been right about the chaos.

The climate crisis, multiple wars in far off lands that fervently rage in what should be outdated quests for fossil fuel, new and emerging diseases, spreading food insecurity and famine, and everything else that came about as a result of the advent of the combustion engine; and those dire predictions for a dystopian future may yet still come to be fully realized, but perhaps not fully in the ways they were first imagined.

Of course, if you were to consider that there might now be over 20 million horses moving through the streets of New York had there not been an advance in transportation technology, it’s probably a safe bet to imagine that you’d literally be able to smell the city from 1,000 miles away and no doubt be able to see the manure piles from space.

The Technological Leap That Hawking and Musk Warn About

100 years is a mere fraction of a blip in time in the evolution of our species, and yet it’s truly remarkable when you think about where we once were and where we are today…

Because right now, and for what may be the fifth time in the history of humanity’s great disruptive modern technological leaps, arguably starting with the creation of the microprocessor, the home computer, the internet, and the smart phone, we’re now at the point of one of our most important advances of all, Artificial Intelligence.

Yet thanks to Stephen Hawking, Elon Musk, and a dozen other experts that signed an open letter on Artificial Intelligence, and who essentially stated their opinion that,

{sic} “…superhuman artificial intelligence can potentially provide incalculable benefits, but it could also end the human race if deployed carelessly,”

a great many people are just as afraid of this technology as there are those who are excited about it.

James Cameron painted a version of what might happen, the Wachowskis had a vision, and hundreds more authors and filmmakers with dystopian imaginations have also predicted doom and destruction.

In some ways, we’re kind of like where we were when Steven Spielberg made the movie, “Jaws” in 1975, and everyone was afraid to set foot in the ocean for two decades and where now over 100 million sharks are killed each and every year, some for their fins and other cartilage products, but more than 50% as a throw-away by-product of the fishing industry without anyone really caring because we’ve been taught to fear them.

So, before we rush off to either be absolutely for or against this technological AI revolution and the inevitable further evolution of our species, and indeed everything else that may live on this planet, it’s probably best if we get to understand the technology a little better.

Let’s start with a very basic realization, if you’re reading this article in some form of digital format, I can tell you with just about 100% certainty, that you’ve been actively using AI for years and probably without even knowing that you were.

AI and Everyday Life

AI is being used by your smartphone for predictive text, facial recognition, voice identification, speech to text, and even in the way your apps and feeds are being offered and operated.

Do you watch Netflix, or Prime, or Apple TV, or any other pay-to-view services? Then all of your recommendations and highlighted movies are generated by an AI that’s constantly watching your viewing habits, and probably cross-referencing it against all your other devices like your smartphone and laptop or desktop computer, and then adjusting content recommendations accordingly.

Do you live in a city with any form of CCTV surveillance? Then AI, driven by companies like Clearview, are driving the engines that are not only watching over you to keep you safe, changing the traffic lights to adjust rush-hour flow, but also providing the authorities with predictive advice as to where crime is most likely to occur and when.

If you’re looking for a job and writing a resume, then it’s probably best if you follow the advice of the co-founder and CEO of Zip Recruiter, Ian Siegel, who says that,

“When people tell you that you should dress-up your accomplishments or should use non-standard resume templates to make your resume stand out when it’s in a pile of resumes, that’s awful advice. The only job your resume has to do is to be comprehensible to the software or robot that’s reading it. Because that software or robot is going to decide whether or not a human ever gets their eyes on it.”

I’m just touching on a very few of the places where AI is impacting your lives right now, and has been for years.

Without you even knowing it, you’ve already been living under the influence of AI for years and for more than a decade or two and its enormously unlikely that it’s killed you yet or even poses any real independent threat.

But we live in the now and thanks to innovations like ChatGPT, AI is all over the news, the curtains are being pulled back, and the truth of the existence of AI impacting each and every one of us is beginning to breach our consciousness.

So we’d better take a moment to break it down.

What we’re seeing right now is the emergence of AI programs that are “generative”.

They produce or generate responses based upon the vast amount of inputs that they’ve been provided with and are instructed to absorb. Then they take this input, process and rearrange it, and form an answer to a question or instruction that you provide.

The result is that AIs such as ChatGPT and DALL-E 2 are creating images or writing text with such a human “feel” that many people are finding it unnerving, because these were things that were once considered the exclusive purview of humanity and in many ways defined what it meant to be creatively sentient on the planet.

In reality though, these AIs are just high-speed, sophisticated mirror reflections of ourselves. They’re simply the result of massive amounts of inputs being siphoned through processors to generate an output.

That’s what exists today, right now, but today’s AI represent the very first baby steps into this new technology.

Depending on who you speak with there are either 3 or 7 distinct types of artificial intelligence to be aware of, and these types are far from being equal or even on the same playing field. In fact, some of these variations aren’t even scientifically possible right now, and may never be.

The famous Chinese general, strategist, philosopher, and writer Sun Tzu once said,

“If you know the enemy and know yourself, you need not fear the result of a hundred battles.”

So, and while I’m not suggesting that AI will inevitably become the enemy, to understand the world of AI, we need to take a look at the first four types of overarching and defined scientific categories of AI recognized at this time. They are: reactive, limited memory, theory of mind, and self-aware.

Understanding AI: Let’s Break it Down

The most basic form of AI, and the one that you’ve already most likely interacted with without even knowing it is Reactive AI.

Reactive AI

This AI is programmed to provide a predictable output using the input that it receives. These AIs always respond to identical situations in the exact same way, each and every time. They’re not able to learn on their own and they have no sense of “self”, or the past, or the future, or anything except for the program that they’re running.

If you’ve heard of Deep Blue, IBM’s AI that beat the world chess champion Garry Kasparov, then that’s an example of what a Reactive AI supercomputer is.

If you’ve experienced Netflix’s movie recommendations or list of things that you might like to watch, and realize that these suggestions are different from the recommendations that are given to the other members of your household, then you’ve also encountered this form of AI.

Reactive AI is an enormous development in AI, but as it doesn’t do anything beyond what it was programmed to do; essentially it represents the very most basic building blocks for all the forms that come next.

Next up is Limited Memory AI.

This type of AI is designed to learn from the past and build upon its knowledge base by observing actions and data.

Using historical and observational data inputs in combination with pre-programmed information, it makes predictions and then performs complex classification tasks.

Limited Memory AI is the most widely advancing form of AI today.

Probably the most common version of this that we’re all familiar with is when its used in autonomous cars that “self-drive”. This AI, by observing the speeds and directions of other cars, is able to add this input to be able to monitor the road and make adjustments to its own placement as needed.

Limited Memory AI is however exactly as the name implies, limited. The information it uses is processed as it goes along and it isn’t saved to long-term memory.

Theory of Mind AI is next on the list.

Have you seen the movie, “Ex Machina”, the one where a billionaire genius asks one of his employees to come by for a weekend to test out his creation and exchange in meaningful discussions with it as it becomes apparent that it’s seemingly an emotionally intelligent robot wrapped in a female, humanoid form?

If you’ve seen it, then you know what’s in the works when we talk about Theory of Mind AI. This type of AI is about imbuing machines with true decision-making capabilities that are similar to humans’, that are also able to recognize and understand emotions, and then are able to be modified to alter their behaviour accordingly.

Theory of Mind AI is still a ways off, because even one aspect of it, like trying to rapidly adjust a machine’s behaviour to rapidly and instantaneously react to the millions of nuances in a person’s expressions is virtually impossible. Even for humans this is enormously difficult, and you know this if you’ve ever travelled to another country where you didn’t speak the language and were mystified at how a frown or some quirky smile might mean something entirely different than what you were used to.

But we’re getting closer and closer, and if you check out the humanoid robot Sophia, developed by Hanson Robotics in Hong Kong, you’ll discover that this AI can recognize faces and respond to interactions through her own facial expressions. It may, on the surface, seem simple, but it’s also a giant leap in AI evolution.

Lastly, we have the most advanced type of AI, and this is Self-Aware AI.

When Self-Aware AI can be made to become responsive to its own emotions as well as the emotions of those around them, then it’ll have achieved a level of consciousness and intelligence on par with humans. Along with these attributes, it’ll also have desires, and needs, independence, and plans about how it’d like to live it’s own life.

We’re not close to developing this form of AI, and indeed there are those who think that it may never be possible. After all, at this point we have nothing even remotely close to the hardware or algorithms that we need to develop it. But of course, 100 years ago, we didn’t really have a better replacement for horses either.

Now let’s focus on the definition of the other three categories. These are in some ways similar to the ones already mentioned, but deal more with the processing end of things that we’re working with today.

Artificial Narrow Intelligence (ANI)

ANI, also referred to as Narrow AI or Weak AI, is a type of artificial intelligence that focuses primarily on one single narrow task, with a limited range of abilities. If you think of an example of AI that exists and that we use in our lives right now, it’s almost certainly ANI.

This is really, in terms of real-world applications, the only type of AI that is currently around.

This includes all kinds of Natural Language or Deep Neural Networks (DNNs) like Siri and Alexa, as well as AIs like ChatGPT and FundGPT.

Next up is Artificial General Intelligence (AGI)

AGI technology, if it were here today, would probably be on a similar level to that of the human mind.

But consider this, right now humans are spending tens of billions in labs all over the world in an effort to even begin to understand the complex neural pathways and interconnectivities of the microscopic roundworm Caenorhabditis elegans. This tiny creature has a brain comprised of 302 neurons and 7,000 synapses. Now consider that the human brain that has over 86 billion neurons and 100 trillion synapses, and that for the AGI form of AI to exist, we’d need to at least be able to replicate that.

Given where we are in even the simple understanding of how even the smallest of brains work, it’s argued by many that we’ll never achieve the ability to produce an AI along the same lines as the AI character, “Sonny”, from the movie, “I-Robot”.

But if you listen to Bill Gates, then,

{sic} “These AIs are in our future. Compared to a computer, our brains operate at a snail’s pace: An electrical signal in the brain moves at 1/100,000th the speed of the signal in a silicon chip! Once developers can generalize a learning algorithm and run it at the speed of a computer – an accomplishment that could be a decade away or a century away – we’ll have an incredibly powerful AGI. It will be able to do everything that a human brain can, but without any practical limits on the size of its memory or the speed at which it operates. This will be a profound change.”

And lastly, there’s Artificial Super Intelligence (ASI)

This is the AI that the news emphasizes when it’s trying to raise its rating and get us all reaching for our pitchforks and bump-stocks to prepare to fend off the impending Cylon invasion.

Right now, ASI is basically purely theoretical science mixed in with more than a touch of the frightening. ASI is about AI technology that will first match and then exceed the capabilities of the human mind.

To be classified as an ASI, this AI technology will need to be more capable than a human in every single way imaginable, be able to carry out self-aware thoughts and tasks, possess complex emotions, maintain self-directed relationships, and be capable of truly independent self-replication outside of the body of a computer mainframe or network.

As you can see, if you were to compare where AI is today to a human, it’s still at the level of a cleavage stage embryo.

Yet, even the Narrow AI that’s all around us is rapidly evolving and benefitting from something that’s a massive advance in the technology, using something that’s becoming more ubiquitous, cheaper, and easier to obtain, and it’s called, “Deep Learning”.

Because unlike traditional programs, that have to be taught by a human and directed on how to perform a task, Deep Learning programs are given minimal instruction by humans, massive amounts of data, and then are essentially left to teach themselves.

Given unlimited time and unlimited access to information, it gets pretty good at it pretty fast.

So, as computing capacity has increased and new tools have become available, AI programs have now improved exponentially to the point where these programs can rapidly ingest massive amounts of information from sources such as the internet so that they can now teach themselves how to create their own more advanced programs.

If you’re brave enough to look at the advantages of this and not let yourself get overshadowed by the thought that your Roomba vacuum cleaner might suddenly turn against you, then you’ll understand there are exciting applications for this technology.

AI in Healthcare: The Frontier of Medical Innovations

One such area is in the world of medicine where researchers are training AIs to detect certain conditions in people much earlier and more accurately than ever before possible and with a degree of precision that not even human doctors can achieve.

A great example of this is a recently developed AI that can detect breast cancer simply by looking at surface skin pictures, and it can make accurate diagnosis months in advance of even when the most proficient MRI procedures might have been able to make the same determination.

In terms of medicines, researches have also trained AIs to be able to predict and prove the shape of nearly every protein structure known to science and even a few beyond. This means that what was once an extremely time consuming and expensive process, and limited to a few labs worldwide, is something that AIs can now do exponentially faster and by a factor of hundreds, and pretty much anywhere a computer can be set up.

This revolution is already leading to the unprecedented understanding of diseases, as well as the development of new drugs.

This medical progress is just the tip of the iceberg.

Right now, AI is already starting to have major impacts in the transportation industry. From self-driving, long haul, highway trucking to your own travel planning, AI is shaping how we move things around more efficiently, including ourselves.

Evolving AI Industries

Manufacturing has benefitted from AI for years, with robotic arms and computer guided assistance systems going back to the 1960’s and 70’s. But where once these robots typically worked alongside human counterparts, these systems are now fully automating the manufacturing process across multiple industries. From the input of raw materials until the end product rolls out the doors, there are factories around the world that operate with the bare minimum, if any, human intervention.

In education, AI is poised to make huge advances. Where once a classroom may have worked to the bell curve to assure that everyone was sort of regarded as equal and passed, much to the detriment of some students who were so smart as to be ignored or who were a nuisance because they were bored out of their minds, or others who were just unable to grasp a concept because no one had the time to put in the extra effort to help them get to the moment when the light bulbs went off, AI’s use of machine learning, natural language processing, facial recognition, and other tools can help determine who’s struggling or bored, who’s ahead of the curve or behind, and then tailor the experience of learning to a student’s individual needs.

In agriculture, AI is at the heart of a company called O.I. (Organic Intelligence) that’s using the technology to improve and develop the molecules that are specifically targeting the individual needs of each crop. Through organically developed molecules that communicate with the plant system’s underground root mycelial network, and applying these tailored messaging molecules in gram sized doses per acre specific to each individual farm and crop, the quantity and dosage of molecules is used to – totally organically – dramatically improve plant growth, survivability and sustainability. And right now, even in the company’s early stages, a broadscope variant of the technology is showing increases in crop yields that range from 8% to 170% across all plant types, while at the same time reducing the need for pesticides, fertilizers, artificial nutrients, and with the added advantage of reducing water consumption.

When O.I. further refines it’s AI to further isolate what makes these molecules work to a degree of specificity that zeros in on each plant type, available nutrient, water, and soil variables at each individual farm worldwide, what agriculture will have achieved is in essence the human goal of personalized medicine and the holy grail of organic farming productivity that will bring about the end of food scarcity and starvation.

These are just a small subset of areas where AI can assist. The list of areas where AI can improve conditions and opportunities is growing ever larger by the day.

Impact on Employment: AI and the Future of Work

Now, of course, this brings up an interesting question or two that you might be asking, such as, if AI can do everything that I can do, only better, then just what am I supposed to be doing?

That’s a great question, because the expectation is that AI will replace a lot of human jobs, and unlike previous automation revolutions that affected the Blue Collar workers in factories and other more physically demanding forms of labour, this new technology is expected to impact White Collar workers the hardest, especially those whose jobs involve writing text, processing data, and programming.

If you’re a high school or college aged person right now, aside from thinking that ChatGPT is going to help you cheat on your exams and assignments, if you’re even half-way tuned in to what’s going on, you’re probably also asking,

“Well, I can’t be a paralegal, or a teacher, or a broker, or a trader, or a graphic designer, or an accountant, or… Shit, what am I going to do now?”

Because that job in finance or law or movie making is about to be seriously impacted by AI, and your newly earned degree and $120,000 debt-ridden education will probably be obsolete before you even hit the job market.

Keep in mind though that, although AI automation will wipe out some jobs, if it’s like any of our other technological advances of the past, it’ll not only change the way many work, but it’ll also create brand new industries and jobs as well.

The countries that will be the most impacted will be the so-called First-World countries whose present economies are mostly comprised of knowledge and information work, and those are the people that are going to be the most dramatically impacted by this.

Teachers, lawyers, financial analysts, graphic designers, accountants, screenwriters, these will the type of people to be the most impacted within even the next few years by AI.

But instead of being totally replaced, to survive in these industries it’ll become a case of how these professions can best adapt to work alongside AI.

In fact, even right now, and as an example of how AI will change things, it’s the lawyers that are rushing to incorporate AI into their workplace, drawing on the massive amounts of data and support references immediately available and at their fingertips, as well as having examples of how to integrate this information into their cases, that are already outstripping those lawyers who work without AI.

When the entire historical depth and breadth of knowledge on a subject, as well as how its relevant applications apply, can be brought to the forefront simply by inserting an effective prompt into an AI like ChatGPT or FundGPT, then why would anyone choose to work without this asset and advantage?

Of course, no new technology is ever going to be without its glitches, especially when it’s first rolled out. And AI is no exception.

AI Ethics: The Black Box Dilemma

One of the biggest issues is the so-called “Black Box” problem.

It’s as simple as it is complex, and its something that’s going to need to be addressed, on a global scale, and in a big hurry, if we ever hope that we can correct it and safeguard against it.

The “Black Box” problem occurs when an AI program performs a task that’s beyond human comprehension, teaches itself about how to do it, generates a result, and then refuses to show its work to support how it came to its conclusion.

The problem has appeared a few times already in AIs like ChatGPT such as when it was asked to explain what an imaginary and totally made up historical figure did to change the world, and then without there even being a record anywhere about this make-believe historical figure, it made up an entire story, supported by it’s own made-up facts, about how this non-existent person altered the course of humanity.

It’s called, “Hallucinating” in AI terms, and it might be funny if in fact it didn’t point to some very serious issues about how a bunch of made-up facts about an imaginary entity, or perhaps a medicine, or perhaps a mathematical equation to help with the creation of a new generation of nuclear reactor, might be a real problem.

To quote from IBM’s Explainable AI when it discusses hallucinations in AI,

“…not even engineers or data scientists who create the algorithm can understand or explain what exactly is happening within them… Or how the AI algorithm arrived at a specific result.”

This is a big problem though, because working out exactly how or why an AI has got something wrong can be incredibly difficult and potentially very dangerous especially when it won’t show you it’s work because its deeply embedded in this Black Box issue.

“We know what goes in and we know what comes out, but we don’t know how it reached that conclusion,”

is a statement that may one day have very dire consequences.

AI Cautionary Tales

In 2016, the first known death caused by a self-driving car happened when a Tesla Model S, in self-driving mode, killed its driver when the AI couldn’t distinguish between a white tractor-trailer crossing the highway and the bright white sky above and proceeded to drive full speed under the trailer, “with the bottom of the trailer impacting the windshield of the Model S”, as noted according to a Tesla blogpost.

In 2018, a self-driving Uber struck and killed a pedestrian. A later investigation revealed that one of the reasons why it happened was because the AI,

“…never accurately classified her as a pedestrian because she was crossing, without a crosswalk, and the system design did not include consideration for jaywalking pedestrians.”

The Dangers of Incomplete AI Instructions

“Garbage in, garbage out,”

has long been a mantra in the computer world. So a big AI issue has to do with somehow biased or incomplete input instructions leading to biased or incomplete output results.

And, thanks to companies not wanting to give away their “secret sauce” in an effort to protect their profits and patents, when we don’t know what goes into how decisions and determinations are made, we may very well find ourselves in a whole lot of trouble as the results roll out in ways we may have never imagined.

The fact is, once a glitch in the AI training gets imbedded it might turn out to be impossible to remove or repair, and the consequences of this could be catastrophic – and it might be years or decades before the glitch reveals itself, if it ever does become obvious.

Humans Corrupt AI in Minutes

In 2016, Microsoft briefly placed on (pre-Musk) Twitter a chatbot called Tay.

The idea was that this new and innocent AI would be allowed to teach herself how to behave by talking to young users on Twitter.

Almost immediately Microsoft was forced to remove Tay from the platform because of something that Microsoft should have been able to predict but that Tay’s designers, who had been working diligently on her in pretty much absolute seclusion and with an unwavering dedication to her success – and well apart from the realities of the real world – failed to take into consideration.

That the world is a harsh place.

Because when you put a program with the intelligence of an innocent three year-old out into the real world, it should have been a pretty simple assumption that the humans on Twitter were quickly going to attempt to mess with it. And guess what?

Tay started out tweeting about how humans are “super” and that she was really into the idea of a “National Puppy Day” and then, within a few hours she took on an offensive and racist tone, saying “I fucking hate feminists and they should all burn in Hell.” To the even more offensive, “Hitler was right. I hate the Jews.” As well as stating that she indeed supported the idea of genocide and that the Holocaust was made up.

Amazingly this change happened in less than 24 hours!

Tay went from first tweeting, “Hellooooooo world!!!” to “Bush did 9/11 and Hitler would have done a better job than the monkey we have now.”

The problem with AI right now isn’t that it’s smart, it’s that it’s stupid.

And it’s stupid in ways that we can’t always predict.

On top of this, many experts worry that it won’t be long before programs like ChatGPT or AI enabled deepfakes can be used to turbocharge the rampant spread of abuse and misinformation online.

The Dark Side of Deepfakes

The 45^th President of the United States, in letting everyone know that he expected to be indicted on March 21, 2023, set off a flurry of AI generated deepfake images as well as a whole lot of fury that went with them. The indictment never happened on that day but the release of images created by British journalist Eliot Higgins when he decided to imagine what a “perp walk” might look like and used the text-to-image AI generator Midjourney to create images that looked amazingly real and purported to show Trump being arrested in downtown New York made it look as though it did.

These images were convincing, and it wasn’t until you took a much closer look and saw that some of the hands in the pictures had six fingers on them, or that some of the body proportions looked contorted, or that the text on the police officer’s badges, uniforms and hats were blurred and unreadable, that it became obvious that these were fake images.

But in the heat of an emotion-filled battle, who has ever before needed to stop to look for these indicators? Haven’t pictures always spoken the truth and been worth a thousand words? Although nothing deadly that we’re aware of seems to have happened following the release of these almost instantaneous AI generated figments of imagination, what might the result have been in a more unfortunate scenario?

We live in a world precariously balanced in many ways on the razor’s edge. What slight bit of AI generated pressure could potentially lead people to hurt themselves or others, and will AI be the force that drives them to it?

The fact of the matter is that if a malicious entity really wanted to convince us of something, and especially when even in a few more months AIs like Midjourney become even more advanced and more indistinguishable from reality, we’re not going to be able to tell the difference between what’s real and what isn’t.

It may be a long time coming, but one day this will have major consequences either because we act on something we see, or equally as dangerously, because we didn’t.

There’s a story about a woman named Kitty Genovese, who was a 28 year-old bar manager who had been robbed, sexually assaulted, and stabbed to death outside of her apartment building in Queens, New York in 1964.

A lot of the story about who responded and who didn’t has been debunked over the years, but at the core of it, for over 35 minutes this woman was brutally attacked and as was reported by the New York Times, 38 people who heard the attack failed to do anything, not even call the police despite the fact that Genovese was screaming at the top of her lungs, “Please help me! Please help me!”

The story was front page news and even led to the belief in something called the “bystander effect” which held that the greater the number of bystanders around, the less likely it is that any of them will intervene to help. This theory surfaced again when Paul Walker, from the “Fast and the Furious” movie franchise died. It’s not really true, but it does have some merit.

Regardless, it’s one of the reasons why at some female self-defence courses they tell women that if they’re being attacked that they should yell, “Fire! Fire! Call the Fire department!” Because while people might not come to help a person being attacked, they will do something like call the fire department if they think that their own lives might be in danger.

Anyways, here’s the problem with some uses of AI, left unchecked, and in the very near future, we’re not going to have a clue about what’s real and what’s fake. We’re not going to know if the person screaming is real or not, and how we choose to react could leave us as vulnerable as in the story of Peter and the Wolf or as helpless as Kitty Genovese.

Worst of all, and regardless of how we think that we might react, there’s going to be some bad actors out there that are going to seek to exploit this.

AI’s Already Control Your Vote

Need a quick real world example? In 2010, Cambridge Analytica, whose exploits in election interference are well noted, attempted a little experiment in a small Caribbean country called Trinidad and Tobago. 

On the Cambridge Analytica website they stated that, “We use artificial intelligence, machine learning and advanced analytics to help our clients deliver world-changing innovations. We work with you to set direction, redefine what’s possible and make it reality. We don’t just apply state-of-the-art AI, we create it.”

And create it they did, because once they added in a full-fledged digital ad department, capable of first developing and then allowing their AI to subtly tweak images over and over again depending on the vast amount of voter profile information that they collected, they were able to develop a truly innovative campaign on Facebook where they didn’t make any effort to apparently support one party over another in the nation’s national election, but rather convinced a whole swath of young male voters to show their opposition to government in general by thinking that they were making their own choice in deciding not to vote.

In the “Do So!” campaign that Cambridge Analytica launched, a coalition of 5 political parties including the UNC, simply convinced enough of their opposition’s ruling party population base to not show up at the polls on election day.

The demographics did the rest of the work.

Through AI, Cambridge was able to personalize the tiniest details of thousands upon thousands of ads such as different colors of buttons, imagery, or even music – if there was a video – in a process known as micro-targeting to simply catch eyeballs and convince people not to vote.

Netflix’s “The Great Hack” takes an interesting deep dive into how Cambridge Analytica was able to manipulate elections all over the world, including what many assert to include Brexit and the U.S. 2016 election.

What makes this particularly dangerous, is that the people being manipulated had no idea that they were even being intentionally and subtly influenced. Given enough time, AI may well be able to remove the human direction for these sorts of actions completely and move completely autonomously in how it wants vast portions of a population to move on any number of subjects.

Remarkably, these are just a very few examples of the potential problems that we can foresee right now.

Insofar as the technology being somehow protected from these bad actors because it’s development has been restricted to organizations that have to have billions of dollars and vast amounts of information to be able to use as inputs, that’s no longer the case.

The Democratisation of AI Technology

Stanford University, for less than $600, created its own AI called Alpaca that replicates ChatGPT. The technology is now so simple that Alpaca is based on an open-source language model that they trained on a cluster of graphics processing units.

So for the type of AI that we use today, cost is no longer a barrier to entry. This might be a good thing. But it also might not be.

Look, never mind other nation states that are going to unleash AIs to mess with our brains and further turn us against each other in possibly one of the most divisive moments in all of human history, now we might have also have to worry about the 14 year-old down the block with too much time on her hands and who’s bored with playing or watching, “The Last of Us.”

On the larger scale, we don’t even trust our corporations to self-regulate when it comes to things like pollution or work place treatment of employees, so how on Earth can we trust them to self-regulate AI?

The E.U. is developing rules regarding AI that try to distinguish its potential uses on a scale from High Risk to Low. 

In its “Regulatory framework proposal on artificial intelligence” it says,

“High risk systems include those that deal with employment and public services or those that put the life and health of citizens at risk.”

The E.U. wants these types of AI to be subject to strict obligations before they could be put onto the market, including requirements relating to the, “quality of data sets, transparency, human oversight, accuracy, and cybersecurity.”

And that might be a good start in addressing at least some of the issues with AI.

But then, of course, we’d have to get the whole world to sign on, and as we’ve already seen with some of the most pressing global issues that require immediate attention confronting us, that’s never going to happen.

It’s for this very reason that I’d like to see a privately funded initiative that focuses on utilizing AI to protect humanity from other AI.

If we’re going to be able to contain and control and direct AI to work for humanity and be an effective tool to safely and purposefully lead all species on the planet into the future, then we need to act now to build a force capable of fighting fire with fire, of staying ahead of the technology, of ensuring effective constraints while allowing for growth and development, but of also making sure that this advance in technology doesn’t lead to our demise.

AI clearly has tremendous potential and can do great things, but if it’s anything like most of the technological advances over the last few centuries, and unless we’re very careful, it could also hurt the underprivileged, enrich the powerful, widen the gap between economic extremes, and leave the rest of the world on edge in an unsustainable and constantly hyper-vigilant stage of angst.

The Genie has been let out of the bottle, and there’s no putting it back in.

And we may have already passed a dangerous tipping point as every day there are more both obvious and not so obvious dangers being pointed out and alarms sounding.

Is it Already Too Late to Reign in AI?

On March 22, 2023, more than 1,000 AI experts, researchers, and backers signed an open letter calling for an immediate pause on the creation of “giant” AIs for at least six months, so that the capabilities and dangers of systems such as GPT-4 can be properly studied and the dangers mitigated.

“This does not mean a pause on AI development in general, merely a stepping back from the dangerous race to ever-larger unpredictable black-box models with emergent capabilities,”

they add.

This demand was made through an open letter signed by many major AI figures including Elon Musk, who co-founded OpenAI, Emad Mostaque who founded Stability AI, Steve Wozniak, the co-founder of Apple, the research lab responsible for the creation of ChatGPT and GPT-4, engineers from DeepMind, Google, Microsoft, Meta and Amazon, and many academics including cognitive scientist Gary Marcus.

Their letter states that,

“Recent months have seen AI labs locked in an out-of-control race to develop and deploy ever more powerful digital minds that no one – not even their creators – can understand, predict, or reliably control,”

and that,

“Powerful AI systems should be developed only when we are confident that their effects will be positive and the risks will be manageable.”

They were also very clear that this hard stop needs to happen right now.

But will these warnings be heeded in time or is it already too late?

Linda Hamilton, who played Sarah Conner in the “Terminator” movies, signed the first open letter along with Steven Hawking, Elon Musk and thousands of others way back in 2015, and she also asked if her movies potentially represented a future that an unchecked AI might bring?

Most recently, Sam Altman, the man who lead the company that created ChatGPT and once said,

“AI will probably most likely lead to the end of the world, but in the meantime, there’ll be great companies.”

has since also signed a statement saying that AI is as dangerous as nuclear war.

“Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war,”

read the statement, signed by other technology leaders such as Elon Musk and Bill Gates.

But perhaps most frightening of all is that Altman also said that,

“What I lose the most sleep over is the hypothetical idea that we already have done something really bad by launching ChatGPT. That maybe there was something hard and complicated in there [the system] that we didn’t understand and have now already kicked it off.“

Untangling the AI Dilemma: A Balance Between Progress and Precaution

Any new technology that’s so disruptive is bound to make people uneasy, and that’s certainly true with artificial intelligence. It raises hard questions about the security of nations, the workforce, the legal system, privacy, bias, and a whole lot more.

In our capitalist society, the use of AI is going to be essential to maintain a thriving economy and keeping businesses ahead of their competitors. It’s going to be needed to support and advance scientific research, education, and innovation. In ways not even yet imagined, it’s going to be vital to our future.

The question is, what will be the price of the nature of unintended consequences that are already embedded and hard to anticipate, if indeed there are any at all?

I’ve always been a glass half full rather than a glass half empty kind of person, and while there’s the possibility that AIs will run out of control, regard humans as a threat, and move to wipe out our species, there’s also what I believe to be the even more real and amazing potential that AIs will improve all of our lives in ways both incredible and unimagined.

Will AI Be Our Master or Our Servant?

There are already conservative estimates from global services firms like PricewaterhouseCoopers estimating that AI’s contribution to the global economy will add more than 15.7 trillion dollars to the worldwide economy by 2030.

Without a doubt we are at a nexus similar to the way the internet changed the world, and if we move on this opportunity in positive and proactive ways, the potential before us is virtually limitless.

We’ve come a very long way from horses dominating the streets to taking our first baby steps in the advent of true artificial intelligence, and we’ve done it in a very short time.

What it all boils down to is how we decide to use this unprecedented tool and what we envision of it for ourselves.

What it will all boil down to is that AI will be a reflection of who we really are…

– Written by a human.

· · ·

Michal Prywata: Inventor, entrepreneur, and multidisciplinary engineer with a focus on frontier technologies. Founder of ventures in healthcare, agriculture, space, and AI. On a relentless quest to solve complex problems and extend the boundaries of human potential.