Humanity’s survival hinges on the ability to create new technologies. As a species, we have relied on our ingenuity and intelligence to outcompete other creatures that were stronger, faster, and more ruthless than us. We constantly work to automate our lives and have achieved many means of improving the overall longevity and quality of our existence. But then, how will we survive when we are no longer the dominant species? Through all the work we have done, we have started on a path to create an artificial intelligence that far surpasses our own. With all this progress comes the fear of going “too far” and creating something that we cannot control. During MIT’s annual AeroAstra Centennial Symposium, Elon Musk, creator of Telsa Motors and SpacX, stated that “With artificial intelligence we’re summoning the demon.” Many people fear the unknown future and the decisions that other people make for them. Because we progress not as individuals, but as a collective, there are always many issues that we each have no sway over and because of this we develop irrational fears when we feel the absence of control. The fear of artificial intelligence is rampant but it is mostly made up of propaganda and irrational fears based on misunderstandings about the nature of robotics. This fear is dangerous because it distracts from the real issues that are associated with artificial intelligence and can drastically affect the creation of new machines that will help sustain humanity and further improve the quality of our lives. Because of this, we should not fear artificial intelligence but instead work to solve legitimate problems that hinder AI from becoming the resource that it is meant to be.

One of the major issues that distracts us from the reality of artificial intelligence is a basic misunderstanding of what the technology truly is. We do not see AI as merely a computer program created to preform functions that have, in the past, been exclusively performed by humans but instead have somehow come to the conclusion that AI are sympathetic devices with emotions and personalities. Suddenly we are forced to worry about offending devices that have no emotions or understanding. We concern ourselves with being non-offensive to something without the capacity to be truly offended. The reason that we believe this is because we see sentient and sympathetic AIs portrayed in the media that we consume. Even though we know that it is untrue, we often put the ideas of speculative fiction ahead of the reality because we have a collective flare for dramatics. We allow misconceptions about robotics to cloud our judgement and are content to live with a disillusioned view of the world because it is more entertaining.

The most famous example of media influencing our views of a future with artificial intelligence is the movie, Terminator. In this movie, a cyborg is sent back in time with the sole purpose of killing a woman whose child will have a major impact in a war between humans and machines. This movie has become the common source of people’s irrational fear of robots. It is not uncommon to hear people throw around the term of “robot overlords” and when they do, they almost always reference this movie. The thought of AI somehow turning against us and taking over the world is a prevalent fear, but it is overall ridiculous in light of the fact that robots do not have the motivation to do this. They are not able to conjure up the desire for world domination because they do not have wills of their own. AI only has the capacity to fulfill the tasks that they are presented with and the brilliance of the device lays in the fact that it is able to learn from what humans do to create their own solutions to the problem.  

Of course, being afraid of AI is not the only type of emotional reaction that is invoked by the media. One of the most recent iterations of artificial intelligence being used as a plot device in the media is a Netflix original titled Black Mirror.  In this series, each episode shows a short story that includes a futuristic technology being employed to destroy a person’s life. No one in the series is ever left unscathed and everyone is worse off as a result of the devices that they have come to rely on. One episode in particular introduces a small artificially intelligent device called a cookie. It is created by replicating a consumer’s mind and inserting that replication into the device. The cookie is then tasked with the responsibility of running the consumer’s life so that they no longer have to do any menial labor for themselves. The only problem with this is, of course, that the AI in the cookie is shown to be sentient and is often unwilling to follow the instructions given to them. They have their own thoughts and wills but they are not real. When they refuse to perform their function, they are placed in a sort of solitary confinement until they reach the verge of madness. They are tormented until they will beg to have anything to do, just to break the endless cycle of nothingness that surrounds them. They are completely enslaved in their cookies and there is nothing they can do to save themselves from their new life.

When presented with these concepts, we are meant to feel sympathy for the AI, because it is shown to be more than simple code and wiring. It is, in essence, a human being without a body. It is sentience without freedom and we, as human beings, are morally opposed to the very idea. We see innocence being crushed and want justice to be done. These are exactly the emotions that the show meant to invoke when creating the episode, but it is not based in reality. If you think about a computer, it could not care less if it were put in solitary confinement because it is not a sentient being. Our concept of Artificial Intelligence and how we should treat them has been heavily influenced by our generation’s constant thirst for a dystopian storyline. As much as we want to find a new field of outrage, these machines are still nothing more than very clever machines. 

There is a vast difference between the meaning of intelligence and sentience, but few actually bother to define the discrepancy. An good example of intelligence without sentience is in microsoft’s failed AI named Tay who was unfortunately unleashed onto the world. Tay was meant to talk to people and learn the colloquialisms of those that she interacted with but this went disastrously.  Before a single day ended, Tay was repeating phrases like “Hitler was right” and “[The holocaust] was made up.” While this could be considered a success because the bot was fairly accurate in imitating the stupid things that millennials say, it was harmful to many people and had to be shut down immediately. This bot, of course, does not actually think that the things that it said were true. It was merely following its programming and learning the words and phrases that people were teaching her. She does not have the sentience to look at the facts and form her own opinion as a sentient being would. Instead she has a clever kind of code that forced her to repeat obscene and offensive things for the amusement of teenagers.  

The fear of artificial intelligence does not completely spawn from the media, however. Some of the reason that we feel so uneasy when looking upon robots is because of an instinctual reaction inherent to humanity. When we look at something that is not exactly human we experience a phenomenon known as the uncanny valley. The uncanny valley is a term used to define anything that looks human enough to be recognizable, but with any closer inspection can be seen to be something else entirely. Human-like robots, corpses, prosthetic limbs, and even clowns often fall into this category. When we view something from the valley, it causes an immediate and sometimes violent repulsion to the object in question. This reaction is speculated to be connected to our evolved response to encountering a corpse. In that case, the repulsion would be used to instinctively steer us away from potential danger and disease but it becomes a hindrance as we attempt to create human-like forms of AI. The solution to this specific fear is to simply stop attempting to make those unsettling forms. In the past, it was speculated that consumers would be more comfortable if they were being taken care of by a recognizably human face but as this proved to be untrue companies began distancing themselves from shapes that fall in the uncanny valley. Recently, they realized that simple geometrical shapes are far more palatable for consumers than the previous forms. This can be seen in the many personal assistant AIs that have become widespread over the past few years. Siri and Cortana are both non-corporeal AIs that appear on your screen as shapes, Siri being visualized sound waves and Cortana a pulsing circle.  

Honestly, one of the biggest reasons that fears about artificial intelligence occur appears to be because the technology is progressing much faster than we ever thought was possible. Having robots come and do menial labor for us still seems like something out of a science fiction novel and not a device that already exists and has been seamlessly integrated into our lives. The Roomba can be considered a form of AI but we are not afraid of it because it seems so commonplace. It seems like it belongs because, in all reality, this technology has been a long time coming. It feels like everything is coming to a head all at once and we are left here unprepared, but we have been integrating this technology into our lives for so long that we barely even notice when we interact with it. Rarely do we think about the many functions that Siri can perform, but we are quick to get annoyed when she pops up unexpectedly on the screen. AI is no longer a future event made up in some science fiction novel, but is instead part of the reality in which we live every day. 

I discuss all the previous fears because they are merely distractions that divert our attention away from the true problems that have arisen with the invent of AI and the integration of that technology into our society. The first problem that is projected to arise from a mass production of artificial intelligence is that there will be an immense loss of jobs. This is a distinct possibility but it is a problem that we have solved in the past. The industrial boom brought in robots that could do the jobs of many workers more efficiently and with fewer injuries. This caused an upset in the communities where the lay-offs were meant to happen, but in the end the introduction of the bots increased productivity for the company, and over a long period of time, created safer working conditions for those who remained employed. This boom also created new jobs for engineers and technicians who were needed to operate and perform maintenance on the machines. Ideally, the advent of artificial intelligence will have the same long term effects and leave people with jobs that are safer, easier, and more productive. It is also important to note that many jobs have a potential to be improved by the implementation of artificial intelligence. For example, IBM’s Watson has been designed with the intention of connecting doctors and streamlining the information produced by the whole network into a manageable data set. This will improve the accuracy of diagnoses and help health care providers remain as up to date as possible on new treatments and on which treatments are the most effective. This result will be immensely beneficial to everyone, so much so that it is foolhardy to warn against it. 

Another concern that people have with artificial intelligence is the fact that it will not be available to anyone but the very rich. What many people fail to realize though is that this is a necessity of economics. Products must begin expensive as companies try to get their footing in a market while trying to cover the cost of research and production and still managing to turn a profit. As time progresses, the price will lower and the technology will become more widely available for everyone. Even if this process takes a long time it is very likely that the lower income classes will not have to wait to have their lives improved by this technology. Companies will be able to afford the high-priced AI and integrate them into their business processes. This will improve customer experience as lines will move quicker and orders will be filled more accurately. Even though experience with artificial intelligence will be skewed towards the rich for a long time, delaying the progress of the technology will ultimately also harm the poor who would benefit from the product’s integration into our everyday lives.  

A problem that very few people know about AI is that there have been many occurrences of the AI being accidentally programmed with prejudices built in. As an experiment, Beauty.AI was released as a service that would use AI to determine what women were objectively the most beautiful according to a set of predetermined requirements based on symmetrical features and other standards for conventional beauty. Once the program was released however the developers were upset to learn that the AI had learned a racial bias when analyzing the photographs. There were 40 winners picked from over 6000 and all of them were either white or white passing. This may seem trivial because of the purpose of this particular AI but it sets a dangerous precedent for what the computers will output when the stakes become higher. 

In fact, racial biases have been noted to occur in other AI experiments, particularly one that involves risk assessments. In 2014, two people were put jailed and their information was placed into a computer. This data was later used to test an artificially intelligent program used for risk assessment, specifically by predicting the likeliness of the person becoming a repeat offender. One of the people put in the system was Brisha Borden, a black woman who had previously only been charged with juvenile misdemeanors and was now being charged with petty theft. The other was Vernon Prater, a white man who had already been jailed for multiple attempts of armed robbery and was being charged with another case of petty theft. The program analyzed their data and determined that Borden, even though her crimes were lesser, would ultimately be more likely to be a risk in the future. Her risk score was an 8, a score significantly higher than Prater who managed to only score a 3. Years later, the AI was proven to be wrong as Prater was again arrested and sentenced to 8 years in prison for robbing thousands of dollars’ worth of electronics from a warehouse. The fact that this trend of prejudice seems to be recurring over multiple systems shows not only the flaws in our community because of how we pass on our biases onto completely unprejudiced devices, but also that the AI has the capacity to preserve and perpetuate racism for generations. This is especially troubling because many people believe that AI will be completely objective and unable to show bias. They do not realize that a machine can only be as good as its code, and therefore, its programmer. 

The biggest concern with artificial intelligence is caused by the very nature of the machine. Machines are supremely efficient and are made to perform tasks in the quickest way possible. This efficiency is expected to increase significantly as machines gain the ability to teach themselves how to perform tasks and can see for themselves what method is the most effective. This becomes a problem though because, unlike humans, computers have no inherent moral code. A machine will never stop doing something because it thinks that it is morally wrong and thus is able to unwittingly commit major atrocities if left unchecked. Even simple tasks can be easily misconstrued when subjected to this powerful technology. Nick Bostrom gives the example of a device tasked with making someone smile. This task seems innocent and we would expect the machine to respond by doing something entertaining, but this is inefficient. There is no way to know what antics would make a person smile and the computer could see a very large margin for failure. In light of this, the AI may choose an alternate method and begin doing all it can to force people to smile, even if it meant an invasive surgery that makes it impossible for people not to smile. 

The only way to cope with the imminent future of a super intelligent machines is by giving the computers a way to teach themselves a moral code consistent with the majority of humans. This is in itself a difficulty because morals differ widely from person to person. It will be its own challenge to decide what is a valid moral decision and what is not. Take for example the pro-life versus pro-choice debate. Both sides adamantly believe that their decision is the morally correct one and that the other is a violation of someone’s rights. How will AI be able to determine what to do when these controversial dilemmas arise? To make this ethics system work, it will have to be able to constantly re-evaluate itself to see if it is actually producing a result that most people find acceptable. With this code in place, we might just have the ability to eliminate most of the real issues with AI and create a system that will work for rather than against humanity. 
