Why AI Is The Most Dangerous Thing You Can Imagine Right Now

Hey Thoughty2 here. Earlier this year Facebook was working on a new highly-intelligent AI chatbot that could talk to and negotiate with humans in a realistic manner. When one Facebook engineer had the bright idea to take two of these AI bots and let them talk to each other, that's when something unexpected and terrifying happened. The two AIs invented their own language that us humans couldn't understand and began using it to talk to each other. Here's a snippet of what the two bots, named Bob and Alice said to each other: Bob: I can i i everything else Alice: balls have zero to me to me to me to me to me to me to me to me to Bob: you i everything else Alice: balls have a ball to me to me to me to me to me to me to me to me The Facebook engineers had no idea what they where talking about, but it was very clear that the AIs did. They had invented their own secret code to converse using the power of artificial neural networks. International news soon reported that Facebook forcefully shut down the experiment due to concerns that the AIs were becoming too smart, getting out of control and to prevent a Skynet situation. This isn't entirely true, yes, they did shut it down right away, but because they were trying to create an AI that could talk to humans and if it was going to take the liberty to invent its own language then that wasn't much use and so the experiment was a failure, in that respect. But nevertheless, this is still really creepy. As we continue to advance Artificial Intelligence technology and push the boundaries further and further every year, what happens when they do become self-aware, sentient beings that are more intelligent than us humans? Could AI take over the world? And will it be in our lifetimes? Artificial Intelligence, the ability for computers to think at a human level and beyond and to self replicate is not a dystopian vision of the future, it is hear, it is now, it is happening. Most people are unaware at the phenomenal rate that our AI systems have improved over the past five years and the trend is only getting steeper and steeper. You may not realise but you use AI every day, we all do. Virtual Assistants like Google Assistant, Siri and Alexa are obvious examples of cutting edge artificial intelligence, but it's also used in many really subtle ways that affect your everyday life. When you see music recommendations on Spotify and film suggestions on Netflix, there is a highly complex computer algorithm behind that, an algorithm that teaches itself with passing time, an AI. This is known as machine learning and large retailers such as Amazon have been using machine learning algorithms for many years now, to predict what you want to buy, perhaps before you even know yourself. In one infamous case in 2014 American retailer Target Stores used artificial intelligence to predict that a teenage girl was pregnant, before any of her family members knew and before she was even sure of it herself. Target's cleverly designed machine learning algorithms had detected that this particular Minnesota teenager's buying habits had changed slightly. Over the past couple of months, she had begun purchasing scent-free soap and extra large bags of cotton wool balls. Upon seeing his daughter's mail one day, the girl's father stormed into his nearest Target store and said the following to the store manager: “My daughter got this in the mail! She’s still in high school, and you’re sending her coupons for baby clothes and cribs? Are you trying to encourage her to get pregnant?” A short while later, the store manager received a phone call from the abashed father in which he said: “I had a talk with my daughter. It turns out there’s been some activities in my house I haven’t been completely aware of. She’s due in August. I owe you an apology.” So to answer the question, is AI smarter than human's yet? The answer is perhaps not all of them, but smarter than one slightly naive Minnesota father, absolutely. Moore's law states that every year the amount of transistors in computer processing units will double. This has held true every single year since Gordon Moore theorised it and the first CPU was invented in 1971. So whilst superhuman artificial intelligence may not be among us just yet, many futurists predict its advent could happen any time within the next 50 years, some say it could be far sooner than that, in as little as 10 years time. Some of the most intelligent individuals in the world who deal with futuristic technology every single day have repeatedly warned that AI could be the most dangerous thing we ever create and our greatest ever existential threat. Stephen Hawkins has made numerous pleas to the global scientific community to tread carefully with AI, or face our own damnation and Elon Musk is constantly expressing his concerns. In an interview Musk recently stated that with artificial intelligence we are "Summoning the devil". Many people who truly understand the power and dangers of Artificial Intelligence have called it the "Last invention we'll ever make - the last challenge we'll ever face". Because when the time comes, if we do get it wrong and we do lose control, there will be a final challenge, a final war, a war against software, a war that we cannot win. Because the AI will be so many millions times more intelligent than the entire human race, so whatever plan we come up with, whatever we try to do to regain power and destroy the AI, the AI will have already thought of well in advance. It would be like ants attempting to wage a war on humans. Except many thousands of times harder. From our perspective, as a species, we feel like we are supremely intelligent, compared to our more primitive animal cousins. But in reality, humans are only fractionally more intelligent than say, an ape. But that tiny fraction of increased intelligence has allowed us to develop tools, language and thus civilisation. If we were to plot the intelligence of every animal on planet earth on a graph from one to one thousand, if mice are at 1, then chimpanzees would be at 4 and dolphins at 8, then that would place humans at a grand total of... 10. Whereas a super intelligent AI would easily be at 1,000, and likely many times higher. Our brains are the most complex organism we have discovered in the entire universe but as marvellous as they are, they are, at the end of the day, limited. They are confined within our skulls, with only limited room for neurons. An AI would have no such limits, our computing power isn't yet fast enough, but when the time comes in the very near future, and it will due to exponential growth, when computer processors are many thousands of times more powerful than they are today, AIs will have access to as much computer processing power as they desire. And since software can travel at the speed of light, all over the globe, there are truly no limits to how powerful an AI can grow. Add on top of that the fact that shortly after a super AIs conception, it will most likely design its own faster, smaller and more transient hardware to run on, hardware more powerful than humans could ever fathom. Before the AIs do take over and your brain becomes irrelevant, you should make the most of it whilst you still can. That's why I regularly use Skillshare to sharpen up my skills, an online learning community with more than 17,000 classes in IT, web development, game design and all kinds of exciting lifestyle, business and creative pursuits. Premium Membership gives you unlimited access to high quality classes from experts working in their fields, so you can improve your skills, unlock new opportunities, and do the work you truly love. I really enjoyed learning about how artificial intelligence can be used to improve peoples lives and real world problems by watching the Skillshare course "Goal-Driven Artificial Intelligence and Machine Learning" If you want to create your own AI then you should definitely give this course a go. Skillshare is also more affordable than most learning platforms out there: an annual subscription is less than $10 a month. But since Skillshare have supported me on this video I am able to offer the first 500 people who click the link in the description, 2 months of Skillshare premium for free, so give it a go today, there's no risk. The very moment in time when humans create such a super intelligence - an artificial intelligence that can re-write its own software and becomes just one percent more intelligent that humans has a name. This theoretical event is called the "Singularity". Such an AI is known within the scientific community as Artificial General Intelligence, all the AI we have created to date is Artificial Narrow Intelligence, or weak AI, since it only operates within specific boundaries and performs pre-designated tasks, such as shopping recommendations. But a general AI would be essentially creating a sentient being, creating new life. Recently the worlds greatest thinkers in the fields of AI and futurism were surveyed on what year they think the singularity will happen, when will we create general AI? The average answer was 2040, just over 20 years from now but some scientists have said it could happen as soon as 2029. Our current technology is not even remotely close, but in only 10 years it's highly likely that it will be, because human progression and innovation is and always has been exponential, it is not linear. In the past 50 years we have made more advances in technology than over the entire history of the human race, and over just the past 10 years, we have made more technological advances than the previous 50. Futurist and chief Google Engineer Ray Kurzweil estimates that by 2021, we will be making more advances and new discoveries in just one year than we have over the entire past century. This is known as the Law of Accelerating Returns. 98% of the scientific community agree that the singularity will definitely happen and most of them believe it will happen in our lifetimes. So what will happen to the human race? Well, the fact is, we don't know and we can never know, until the moment it happens. One thing that all AI researchers agree strongly on is that when the singularity happens one of two extremes will happen - the human race will either become extinct or become immortal. When creating AI we are essentially creating a god, and we can't know if it will be a kind god or an evil god until its too late. This is because the AI will be so many magnitudes smarter than us, so our puny minds can't even begin to imagine what a being of such supreme intelligence would be interested in. What would its desires, hopes, ideologies and morals be? If this omniscient AI turns out to be a merciful god the future will be an incredibly exciting time but we'll look at that in a minute. Let's first take a look at scenario number one, the extinction of the human race. Because, you know, it's nice to get all the doom and gloom out of the way. If evolution has taught us one thing it's that a species will do whatever it can in its power to survive and reproduce. There's no reason why AI would be any different. Our planet contains limited resources and if the AI deems the entire human race as an unnecessary consumer of said precious resources, then it could decide to terminate us to free up all those lovely resources and energy The only reason humans are civil and decent to other humans, most of the time, is because they view them as intelligent beings. Why do we find it so easy and guilt-free to swat a fly yet we wouldn't instantly condemn a stray dog to its death if it found its way into our home? Because the dog is more intelligent than the fly, so humans, for some reason, believe that that particular creature's life is worth more, due to its intelligence. This is called humanisation and its an ability that psychopaths don't possess, their lack of empathy when harming others comes from the fact that they are unable to humanise their peers, they see them as lesser beings. So many Nazis felt comfortable and proud with mercilessly slaughtering millions of Jews, because Hitler had quite wrongly portrayed them as less intelligent and less worthy beings than the German peoples - he had de-humanised them. An AI would be so many magnitudes more intelligent than us that the chance of them de-humanising us and seeing us more as a pest, similar to how we see ants and mosquitos, is very high. This would erase any moral guilt they could potentially have about committing genocide. And how would they go about said genocide? It would be a grand plan so intelligent and complex that we can neither predict it nor see it coming. Something like, the AI creating trillions of nano robots that would enter into the bloodstreams of every human on Earth undetected and then, simultaneously release a tiny quantity of the most toxic poison that the AI is capable of creating, instantly killing over 7 billion people. We have no idea how a super AI would think, it could be because of the most basic oversight in its design. For example, if we create an AI to grow potatoes for us and its primary goal is to grow as many potatoes as possible. It could one day decide that to grow more potatoes it needs more land and humans are taking up that land, so it eradicates all life on Earth, so it can have more room for potato fields. This seems like a stupid prospect, but its entirely possible, because we don't know what the motives would be of an AI, there is no guarantee it would view life with the same esteem as we do. One thing that will not happen is the classic Hollywood image of killer robots out of control. Such "hardware" would be far too primitive and rudimentary for a super AI, it's choice of weaponry would be something we don't even know exists yet and something far more effective and terrifying. But what about the more cheerful outcome? When the singularity hits, what if we discover that our new sentient overlord is merciful and kind, then boy are we in for a wild and exciting future. Once the AI reaches levels of intelligence many thousands or even millions of times that of the human race there are essentially three roles that it could take within our world. It could act as an oracle, doing no more than answering any question we ask it. Such as "How do we solve world hunger?", "How can we build a time travel machine?", "What is the answer to life, the universe and everything?". So it would essentially be Deep Thought. The second role it could undertake would be more like a super intelligent servant, a genie if you like, where it would actually enact what we asked of it. A super-intelligent being would very likely know how to modify, dissemble and re-assemble atoms and even particles, and build anything it likes. So we could ask it to build us a time machine, and it would, using some advanced atomic-level construction technique that us humans couldn't even begin to comprehend. The final role it could take on is a ruler, overlord or sovereign. Basically, a god. When the singularity happens there will be an intelligence explosion, because of the recursive self-improvement that the AI will be capable of, it will shoot past human level intelligence like a rocket. Within days, possibly even hours it will have solved world hunger, cured all diseases and hereditary conditions, reinvented human transportation, possibly even introducing us to teleportation and figured out the most stable and fair political and economic systems to run our world. But all this is only the beginning. When I say we will have created a god, I'm not exaggerating. The AI will effortlessly be able to create and destroy life and shape the very earth itself. Think about it, for a few years now scientists have been able to manipulate and move individual atoms in the lab. If it's possible for us dumb humans to move atoms then it goes without saying that a super-intelligent being could effortlessly rearrange billions of atoms all at once, meaning nothing in the universe will be safe from its touch. The AI god could move mountains and part the seas, Moses-style. Anything and everything could be transmuted into something else. Garbage could be transformed into bags of rice or even chunks of meat, we would no longer need to farm animals because atomically-identical meat and other foods could be created out of anything. This of course leads to immortality. If the AI can re-arrange atoms then it could, theoretically create an anti-aging box that an eighty-year-old could walk into and come out the other side twenty-years-old again. The AI could even create super-intelligent nano bots that would permenantly live inside all of us and would constantly repair damaged cells, so we would no longer age. After all the body is like a car, if we can repair it from the inside it will keep going forever. The human body is just a lot more complex than a car engine, but it would be child's play to a super AI. And if we were to get knocked down by a bus? No problem, we could of course ask the AI to back up our brains in advance and a new body, even a new identical brain, could easily be created and the backup be restored, if the worst were to happen to us. We truly would become immortal. Of course, we would eventually come to the realisation that our bodies and limited brains are only holding us back. We would eventually abandon our physical bodies and upload ourselves into a new digital realm, where we too would become an AI, just as intelligent as the AI god that we created. And once we have all become sentient, space travel would be a breeze, since we would all be software, we could simply beam ourselves on a laser beam to other planets and other galaxies. All I have talked about seems like science fiction, possibly scenarios that could unfold thousands of years in the future, but the terrifying reality is that we are on the very cusp of all this happening within the next fifty to one hundre years or sooner, the majority of the world's greatest scientists agree with this. So the next twenty to fifty years are going to be either terrifying or exciting, but one thing is for sure, we are about to create our very last and the most significant invention in our entire 2.5 million year lifetime as the human race. Thanks for watching.