<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Eric Spector on Medium]]></title>
        <description><![CDATA[Stories by Eric Spector on Medium]]></description>
        <link>https://medium.com/@ericspector?source=rss-e84f3787043c------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*47m50uG3HR093Q7C0f_arg.jpeg</url>
            <title>Stories by Eric Spector on Medium</title>
            <link>https://medium.com/@ericspector?source=rss-e84f3787043c------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 07:49:05 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@ericspector/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[The Ethics of Pain in Artificial Intelligence]]></title>
            <link>https://medium.com/@ericspector/the-ethics-of-pain-in-artificial-intelligence-87447f52240e?source=rss-e84f3787043c------2</link>
            <guid isPermaLink="false">https://medium.com/p/87447f52240e</guid>
            <category><![CDATA[ethics]]></category>
            <category><![CDATA[black-mirror]]></category>
            <category><![CDATA[functionalism]]></category>
            <category><![CDATA[utilitarianism]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Eric Spector]]></dc:creator>
            <pubDate>Thu, 10 May 2018 00:21:39 GMT</pubDate>
            <atom:updated>2018-05-10T00:22:47.822Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*RgnfrONKUsRTHbFvtVlxXA.jpeg" /><figcaption>Jon Hamm in Black Mirror’s White Christmas</figcaption></figure><p>The advent of artificial intelligence (AI) has forced many to question the defining features of consciousness and perception. Can non-biological, silicon-based machines perceive the world as do humans? Can they feel and understand pain? This philosophical theory of multiply realizable mental states is called Functionalism. If this theory is achievable, it has many implications on the future of morality. For instance, some argue that exposing conscious robots to pain can benefit society, while others oppose this notion.</p><p>In this paper, I define what is best for society under the context of <a href="https://plato.stanford.edu/entries/utilitarianism-history/">Utilitarianism</a>, or the doctrine that defines ethical actions as those which impartially promote maximal happiness on the greatest number of beings possible. This in turn includes avoiding unnecessary pain as much as possible.</p><p>With the above-stated concepts of Functionalism and Utilitarianism in mind, I plan on formulating the following argument: If you are a Utilitarian and subscribe to the notion that pain is objectively morally wrong, and you are also a Functionalist and believe that machines that can feel pain can be built, then you must believe that making such machines is objectively morally wrong.</p><h3><strong>Functionalism and its Role Within AI</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/767/1*NQ-vVxmSQCqB-FymFVHAaA.jpeg" /><figcaption>Functionalism: they are ALL chairs!</figcaption></figure><p>According to the theory of <a href="https://www.iep.utm.edu/functism/">Functionalism</a>, mental states are defined by what they do rather than what they are composed of. Consider, for example, chairs. Chairs can be tall, sharp, wooden, sturdy, and square. They can also be short, round, plastic, flimsy, and oblong. The essence of a chair is not dependent on the materials from which it is made, but rather the fact that it provides a place for humans to sit. Similarly, if the job of a mind is to think, feel, intend, etc., these different mental states can be completed by something composed of metal and silicone, not just biological gray matter and neurons. Just as chairs can be multiply realizable, so, too, can mental states.</p><p>In keeping with this concept of multiple realizability, it is becoming more convincing that consciousness through artificial means can soon become a reality. Although hyper intelligence in machines exists at this point only in science fiction, a lot of capital and research has been committed to advancing the technology. Researchers at Leibniz University of Hannover, for example, <a href="https://spectrum.ieee.org/automaton/robotics/robotics-software/researchers-teaching-robots-to-feel-and-react-to-pain">have developed an “artificial robot nervous system to teach robots how to feel pain.”</a> In essence, human feelings of tactile sensations have been replicated in robots, and once stimulated with enough force or heat, these robots have developed reflex-like reactions to pain. There is debate about whether or not these reactions truly represent mental states, though.</p><p>Many, including philosopher Daniel Dennett, subscribe to the notion that robots can simply understand pain rather than truly feel it. In his paper titled <a href="https://www.jstor.org/stable/pdf/20115302.pdf?refreqid=excelsior%3A148914505a70e371ff9e78ce48e809a4"><em>Why You Can’t Make a Computer that Feels Pain</em></a>, Dennett calls on the reader to partake in a thought experiment about a computer simulation of a hurricane. As with any simulation, this computer’s task is to mimic reality. But, to Dennett’s point, there is no expectation to get wet or blown away in the presence of the simulation. The point of the simulation is to imitate, not replicate. The computer may know the physical traits of a real hurricane, but it cannot recreate the actual process. Analogously, a computer may be able to understand pain, but not recreate the actual feeling of it.</p><p>Whether or not the researchers at Leibniz University or Dennett are correct in the reality of the possibilities within AI, let us presume the former for the rest of this paper. In the event that computers <em>are</em> capable of such emotional response to stimuli, we should prioritize mere understanding rather than feeling. How a being processes pain plays an important role in decision making.</p><h3><strong>Feeling vs. Understanding Pain: The Latter is Better</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*VFVV9eh48gUGpYcw9amuUA.jpeg" /><figcaption>Feeling vs. Understanding</figcaption></figure><p>Pain plays an immense psychological role in our lives. As an intricate experience that is methodically tuned by the brain, pain forces us to deviate from certain environments in search of more comfortable ones. Many times, though, the feeling of pain serves as a distraction as opposed to a useful tool.</p><p>Prominent neuroscientist Vilayanur S. Ramachandran has been credited with <a href="https://www.painscience.com/articles/pain-is-weird.php">helping relieve pain from phantom limbs</a>, and has been quoted saying that “pain is an opinion on the organism’s state of health rather than a mere reflective response to an injury.” Feeling pain, therefore, can be understood as an <em>emotional</em> response to harsh physical stimuli. In the absence of this emotion, only an understanding of that stimuli would subsist. In most circumstances, this objective and accurate physical understanding of pain can play a more useful role than that of a feeling. Understanding, over feeling, can help people see the world as it is, not what it seems. Why, then, would anyone want to create AI that can feel pain in the first place?</p><h3><strong>Creating AI that can Feel Pain: Arguments and Rebuttals</strong></h3><p>Under the presumption that pain and suffering within humans is wrong and should be avoided, the same standard should be held for AI as well. For if defending consciousness from pain is the ultimate goal, and consciousness can be multiply realized in AI, then it is in everyone’s best interest to defend against pain in AI. However, there are some who oppose this notion, and, under the guise of maximum <em>human</em> security, defend the idea that pain-receptive AI should exist.</p><p>In the case that AI can feel pain, what benefit would this provide? The following four arguments have been presented to showcase the benefits of pain in AI:</p><p><em>1. Pain has proven to be evolutionarily advantageous<br>2. The more pain AI feels, the less pain human beings feel<br>3. Pain is the pinnacle of consciousness<br>4. Pain is essential to joy</em></p><p><em>Pain has Proved to be Evolutionarily Advantageous<br></em>It is true that the ability to feel pain has played an important role in the survival of species. For example, most invertebrates, when exposed to noxious stimuli, retreat from the source of pain. Many mammals have evolved to cry out in response to extreme pain, which signals to the rest of the pack to provide protection, as an animal in pain is an easier target for predators.</p><p>Although pain has proved to help in the survival of biological species, this does not deem it necessary for the survival of AI, which can be programmed to understand its environment without feeling it. Furthermore, adding a survival defense mechanism may align with human beings’ best interest. In the event of AI’s threat to society, detection of harm by way of noxious stimuli may not be advantageous to human survival if, for some reason, the ultimate goal were to protect AI consciousness above biological consciousness.</p><p><em>The More Pain AI Feels, the Less Pain Human Beings Feel<br></em>Many foresee a future in which robots work alongside humans. The year 2016 in the United States alone <a href="https://www.bls.gov/news.release/archives/osh_11092017.htm">witnessed 892,270 occupational injuries and illnesses</a>. The advent of robotic coworkers would cause the amount of on-the-job human pain to reduce drastically. Researchers at the aforementioned Leibniz University are hoping that the physical pain experienced by humans will transfer to their robotic counterparts, thereby reducing the amount of human suffering in the world.</p><p>The popular science fiction anthology series <em>Black Mirror </em>has also explored the notion of AI labor for the sake of human comfort. In an episode titled <em>White Christmas,</em> a character, Greta, is looking to more personalize her smart home. She has very specific needs and wants maximum efficiency in her operational home. As a result, she undergoes a procedure that copies her consciousness onto a device connected to her home system. This consciousness, or software, retains Greta’s desires, feelings, and understanding of the world so that she can better serve her original counterpart. Predictably, she refuses to act as a slave for Greta’s petty needs, and in turn is subjected to mental torture by the company that created her. Her perception of time is accelerated and she essentially lives in a white room of solitary confinement for six months with no one to speak to and nothing to do. This was all done with the intention to break her willpower so that she will submit to work for the needs of her human counterpart.</p><p>But is this really ethical? If mental states are multiply realizable, there should be no hierarchy of pain infliction. Ensuring the safety and comfort of human beings over conscious AI should be considered just as antiquated as abusive slave labor. Under the context of Utilitarianism, impartiality must persist even among the variety of intelligences.</p><p><em>Pain is the Pinnacle of Consciousness<br></em>Carl Jung has famously said “there is no coming to consciousness without pain.” (Contributions to Analytical Psychology, p. 193). Although he was talking about marriage at the time, many people still hold the notion that pain is essential to consciousness and a useful tool in feeling closer to our humanity. The argument appeals to the notion that <a href="http://www.socrethics.com/Folder2/Biology.htm">one never feels fully alive until he feels the sensation of the combination of intense sensory experience aligned with emotion and a distinct focus on the present</a>. If such experiences play a significant role in our perspective, would it be logical to deprive AI of that same familiarity?</p><p>In rebut against this notion, one has to make an obvious case: if pain played such a naturally important role in our lives, people would seek it out. Nevertheless, we aim to avoid physical pain and advance, instead, towards other emotionally rich experiences, such as joy and excitement. Deep joy, for that matter, can also aid in experiencing the same intense consciousness as described above. Many adventure seekers jump out of airplanes and climb the tallest of mountains to realize these states of hyper consciousness. Perhaps most important in achieving this feeling is the presence of adrenaline.</p><p>Furthermore, the argument suggests that one cannot achieve consciousness without feeling pain. In continuing the aforementioned chair analogy, this would be akin to removing a leg of a chair. Although the experience of sitting may be altered without the fourth leg, the chair still performs its role of allowing a human to sit. Similarly, removing pain from the equation of consciousness alters the subjective reality, but the other processes of the intelligence remain intact and consciousness still persists.</p><p><em>Pain is Essential to Joy<br></em>In relation to joy, many argue that it can only be realized when contrasted with pain. Therefore, pain is necessary for the existence joy. This appeal to the understanding of opposites echoes arguments of the 5th-century theologian philosopher, St. Augustine. Although joy may be better experienced after the experience of pain (i.e. a relaxing shower after a long run), this does not make pain necessary for joy to exist. Similar to the above-stated rebuttal, people do not seek out pain to experience more joy, and it would be illogical if they did. Additionally, even if humans were wired to need pain to experience joy, this programming can be customizable and changed within AI to make that supposed previous necessity irrelevant.</p><h3><strong>Utilitarianism and Decreasing Amount of Suffering Among Beings</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/590/1*hReW18wmUtujprFY86wCjQ.jpeg" /><figcaption>The character of Greta in Black Mirror’s White Christmas after being subjected to torture.</figcaption></figure><p>Using Utilitarianism as a guiding concept towards a better world implies decreasing as many negative mental states as possible. Visualize sitting down at a dinner party. You find your name on your placemat and responsively sit in your chair. You realize that your chair is deficient. One of the legs is shorter than the other, and it is causing the chair to wobble back and forth depending on where you place your body weight. You think to yourself that it will distract you for the rest of the dinner and immediately try to mitigate the situation by temporarily placing a sugar pack underneath the stunted leg.</p><p>A short leg on a chair is akin to feelings of pain in a conscious being. These feelings are a distraction from happiness and do nothing to better the experience of consciousness. In an improved, more utilitarian world, unnecessary feelings of pain are diminished and replaced with other forms of mental states — perhaps happiness, curiosity, intrigue, and excitement.</p><p>In the event that AI truly can feel pain, a utilitarian world holds that they would be better off without such an ability. Increasing the amount of unnecessary pain in the world is an unethical act, and this concept will become increasingly important as advances in robotics and AI come into fruition.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=87447f52240e" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>