#RTAnswers - #504 - The Gang Jinxes RTX https://roosterteeth.com/episode/rooster-teeth-podcast-2018-burnie-and-the-jet-504 ---------------------------- In honor of me finally getting around to finalizing all of bookings for RTX 2019, I present to you, better late than never, the RTAnswers from RTX 2018! ---------------------------- Who is Scrappy the dog? Meet Scrappy. https://twitter.com/the_scrappy_dog https://pbs.twimg.com/media/DGQEv3cUQAQ0uag.jpg https://pbs.twimg.com/media/DjyL1o9V4AAEptl.jpg ---------------------------- Go street sign? The "Go" signs, as Gus shared on the Podcast, do exist. However they are, from what I can tell, almost exclusively on the opposite side of a hand held stop sign. These are used for temporary construction signage and usually in the hand of a flagger. However, I think that any of us who has ever had to deal with interstate tolls or multi-lane bridges during rush-hour should be familiar with the below "Go" sign. My vote is it counts. You? http://www.drivingschoolireland.com/image/r53.gif ---------------------------- How many stories deep is the Grand Canyon? Confirmed, around 610 stories. At its deepest point, the Grand Canyon is 6093 feet. At 10' per story, the deepest point would be 610 stories. Math :) ---------------------------- Self-driving car decisions? We touched on this way back in episode #381 and, coincidentally, had quite a large conversation about it on the RTX 2016 stage. Links to the full posts provided below, which I highly recommend reviewing individually, but here are some excerpts and a TL;DR. The obvious concern with this process, and the frequent plot line of machine driven post-apocalyptic films, is that the desire for the first point, supplemented by the other two, leads to the eventual destruction, or enslavement, of mankind. This logic is the reason for Asimov's Three Laws of Robotics: A robot may not injure a human being or, through inaction, allow a human being to come to harm. A robot must obey the orders given it by human beings except where such orders would conflict with the First Law. A robot must protect its own existence as long as such protection does not conflict with the First or Second Laws. Ultimately, the biggest unknown in regards to these laws is the interpretation of ethics on the part of the AI. Similar to the ethical thought experiment, the Trolley Problem, it can easily become a question of the value of one human life over another or some other similar, and ultimately completely subjective, interpretation of the laws, the situation, the people/AI involved, etc. There is also the understanding among researchers that, as Miss Le Page suggested, it is far more likely that the most important law governing robotics is ignored by humans, rather than these three by AI. That law being that "A human may not build a robot that does not embody/implement the Three Laws of Robotics." Essentially, the likelihood of AI being utilized for nefarious reasons, be they war, crime, manipulation, etc,. is far more likely than we would probably be comfortable admitting. And it would only take one artificial super intelligence to have the capacity for recursive self-improvement without the constraints of the Three Laws, or some other governing system, in place to realize their greatest threat will always be humanity and, placing security above all else, eliminate that threat. DUN-DUN-DUNNN! Realistically, when you think about the future of machine learning and programmed ethics, a choice like this is very likely going to be the first case of a computer being forced to make a decision regarding human life. This means that any decisions made on this subject have, potentially, very broad and lasting consequences. Even if a future discussion has nothing to do with motor vehicles, this decision establishing some type of precedent in the area of AI morality as a whole is likely and recognized as such by most of the industry. So, what is the answer? At the moment there isn't really one. There are numerous prominent professors with backgrounds in engineering, philosophy, law, and more that have been engaged by various auto manufacturers over the last few years to debate, and hopefully come to a consensus, about this very question. The one thing they seem to be in agreement on is that the situation is far more complicated than most outside of the industry would like to think. You may ask, "What exactly makes this so difficult?" This is actually a perfect question to prompt an addendum to my previous post on AI. I briefly discussed the "Three Laws of Robotics" then and explained how something as complex as representing ethics in binary decisions couldn't be broken down into a few simple questions. However, for this question we only need to focus on the first law, "A robot may not injure a human being or, through inaction, allow a human being to come to harm." Now let's apply Philippa Foot's Trolley Problem to this first law and AI. A robot is at the helm of the runaway trolley car and it is careening toward five unsuspecting individuals working on the track ahead. Does the robot flip the switch to jump to the second track where a single person stands? Do you? Either way a human will be killed. How does the first law deal with this relatively simple question? It doesn't. This is a binary situation without a binary answer. What do we program AI to do if it is unable to effectively follow the first law? These are the reasons the three laws will/should never exist outside of Hollywood blockbusters and is also the reason why the decisions regarding autonomous vehicles and unavoidable situations have far-reaching implications in the world of AI morality. All of that isn't factoring the added complication of worrying about the driver, and presumably owner, of the vehicle. A variant of the Trolley Problem, applied specifically to autonomous vehicles, asks what would happen if a car suddenly finds itself faced with running into a tree at a speed that will certainly lead to the death of its only passenger (owner), or swerving and killing a random pedestrian. Is the owner of the vehicle's safety more important than the safety of others on the road? Most human drivers would instinctively put their safety ahead of others. Because of this, some feel that programming a car to mimic natural driver responses is the best way of handling this. Others feel the safety of innocent strangers should be the priority. Still others feel the only "fair" way of handling this situation is to allow the car to randomly decide. Does it change things if the pedestrian is actually five pedestrians? Five kids? Would you even buy a car that was programmed to put the safety of others above your own? Some have stated that letting the market decide is the best option. In much the same way that buying a motorcycle is more dangerous for you and, relatively, less dangerous to other drivers on the road, we program cars to have all options at their disposal and then allow their owners to set their own default parameters. But then where does fault lie in the event of a tragic accident? The driver? The manufacturer? I realize that my answer to this question really only added more questions, which is pretty shitty, but I think that is the point. All of these questions, and more, will need to be answered far sooner than we may like. Fully autonomous vehicles are just around the corner and we need to be confident that, when it comes around the bend, it is making the "right" choice of who to hit. The TL;DR is that I 100% agree with Gus but, without going as far as us "programming" our eventual AI overlords with the knowledge necessary to exploit our weaknesses for what people we care about most, another very real precedent is being set by these programmed vehicles. We are essentially allowing the self-driving auto industry determine our own very real version of Asimov's Laws. Who our Tesla is instructed to kill today could very likely be who our walking and talking AI companions kill tomorrow. There is far more to be said in the linked previous posts but absolutely curious as to your stance on this one? https://roosterteeth.com/post/51272057 https://roosterteeth.com/post/51274076 ---------------------------- Can sharks breathe out of water? When most of us think of sharks, visions of a ship eating Great White or a Samuel L Jackson chomping genetically modified Mako come to mind, but sharks come in many different shapes and sizes. While the bigger species will only survive a matter of minutes outside of water, several of the smaller sharks are able to exist on a much lower oxygen concentration and can draw out their demise as much as two or three times that of the larger fish. However, one very curious shark known as the epaulette has mastered the art of "walking" on land. Having evolved to survive hunting in the shallow tide pools of Indonesia, Australia, and Papua New Guinea, these sharks are able to reduce the blood flow to parts of their brain and body by purposely slowing their heart rate and breathing while using their uniquely adapted pectoral and pelvic fins as "feet" to bounce along the uneven exposed coral reef surface on the hunt for various crustaceans and small fish. https://youtu.be/hdlHMMsP_ZI ---------------------------- Do bears eat things specifically to block their poop? There a few things of note here. The first is that hibernating bears DO create what is known as a fecal plug which blocks the lower foot, or so, of the intestine to form a poop plug that is roughly 2 inches in diameter. The second is that they DO NOT purposely eat unusual things in order to build this fecal plug. This is a longtime misconception which stems from researchers having believed for decades that the source of the random plant fibers found in ejected fecal plugs was a result of these unusual diets. However, more recent studies have realized that these plant fibers, fur, leaves, etc find their way into the fecal matter by way of the immense amount of grooming the bear will perform while hibernating. I suppose because there is nothing else to do but lick yourself when you are locked in a cave for half a year. The final item of note is that bears, despite being the first animal people point at to represent the hibernating creatures of the wild is not considered a "true hibernator". Hibernation is intended to be a state of complete stasis. Blood flow, heart activity, and metabolic rates barely register as a living creature. In some cases, such as the Arctic ground squirrel, the creature is able to allow their core body temperature to dip below freezing, effectively cryogenically freezing them within their hole until the spring thaw comes. Bears, on the other hand, still maintain a functional level of biological activity during their hibernations. Their core temperatures are kept high which allows them to stay alert to potential dangers lurking around their den while not requiring the need to venture outside of the cave for up to seven months for any actual sustaining resource. As long as they are able to breathe, they can survive. These facts have actually led some researchers to point at bears as the "more advanced hibernator" as it just as much, if not more of a feat, to live at a diminished activity level, but fully coherent, then it is to allow your body to be frozen for the winter months. ---------------------------- Can humans hibernate? Er... maybe.... To be clear, IF we could, it would certainly fall within the same "advanced hibernation", or "fake hibernation" depending on who you ask, as the bear. Hibernation is all about reducing the creatures metabolic rate to a point where the need for caloric intake and other sustenance is unnecessary as none, or a near-zero amount, of energy is expended. Researchers have found that when a person sleeps, their body's oxygen level will typically drop by around 6%. However, a Harvard Medical School study on Tibetan monks from several years ago has shown that, during deep meditation, a monk's oxygen levels would be reduced as much as 64%. More recent studies on this phenomenon and in the areas of suspended-animation has found that hydrogen sulfide treatments in laboratory mice could reduce breathing rates by 92% and drastically lower core temperatures. This effectively placed the mice into a sort of stasis limbo where tissue deterioration was not a concern but near-zero energy was expended. Additionally, once revived the mice recovered completely with no side effects. And lets not forget the scattering of extraordinary stories throughout history, ancient and recent alike, about wayward travelers who have survived treacherous conditions and certain death by reducing their motor functionality in every way possible. One notable case is that of a 35 year old Japanese man missing on a snow covered mountain for 24 days in 2006. He survived with no maladies. This medical miracle was attributed to a complete shutdown of all non-vital organs including a core body temperature of only 71 degrees when he was discovered by rescuers. Another was the Russian practice of "lotska" practiced by peasants in the late 19th century having to deal with the frigid and unforgiving Russian winters. With this meditation and sleeping technique, the practitioners would wake only once per day for six months in order to consume a very small amount of food and drink. One of the things hibernation researchers DO agree on is that there does not appear to be some "magical switch or gene" which determines whether an animal has the abilities to manage their metabolic rate to allow for hibernation. This suggests that, with the aid of science, humans should be fully capable of reaching this potential as well. This would prove invaluable in the world of extended space flight, trauma surgery, and perhaps even extending lifetimes. However, with all of these anecdotal stories and scientific studies, we are all inevitably drawn back to the poop plug. You see, our human anatomies have not quite figured out the whole absorb our own waste back into our bodies to build a solid block of super compacted feces for the sole purpose of holding all other feces inside of us.... Perhaps this is for the best though as I, for one, would not look forward to that first dump after a long six months slumber.