{"id":166867,"date":"2021-01-30T18:00:46","date_gmt":"2021-01-30T15:00:46","guid":{"rendered":"https:\/\/en.buradabiliyorum.com\/this-new-book-explores-the-difficulty-of-aligning-ai-with-our-values\/"},"modified":"2021-01-30T18:00:46","modified_gmt":"2021-01-30T15:00:46","slug":"this-new-book-explores-the-difficulty-of-aligning-ai-with-our-values","status":"publish","type":"post","link":"https:\/\/buradabiliyorum.com\/en\/this-new-book-explores-the-difficulty-of-aligning-ai-with-our-values\/","title":{"rendered":"#This new book explores the difficulty of aligning AI with our values"},"content":{"rendered":"<div id=\"ez-toc-container\" class=\"ez-toc-v2_0_85 counter-hierarchy ez-toc-counter ez-toc-custom ez-toc-container-direction\">\n<p class=\"ez-toc-title\" style=\"cursor:inherit\">Table of Contents<\/p>\n<label for=\"ez-toc-cssicon-toggle-item-6a2fcd8a261e0\" class=\"ez-toc-cssicon-toggle-label\"><span class=\"\"><span class=\"eztoc-hide\" style=\"display:none;\">Toggle<\/span><span class=\"ez-toc-icon-toggle-span\"><svg style=\"fill: #dd3333;color:#dd3333\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\" class=\"list-377408\" width=\"20px\" height=\"20px\" viewBox=\"0 0 24 24\" fill=\"none\"><path d=\"M6 6H4v2h2V6zm14 0H8v2h12V6zM4 11h2v2H4v-2zm16 0H8v2h12v-2zM4 16h2v2H4v-2zm16 0H8v2h12v-2z\" fill=\"currentColor\"><\/path><\/svg><svg style=\"fill: #dd3333;color:#dd3333\" class=\"arrow-unsorted-368013\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\" width=\"10px\" height=\"10px\" viewBox=\"0 0 24 24\" version=\"1.2\" baseProfile=\"tiny\"><path d=\"M18.2 9.3l-6.2-6.3-6.2 6.3c-.2.2-.3.4-.3.7s.1.5.3.7c.2.2.4.3.7.3h11c.3 0 .5-.1.7-.3.2-.2.3-.5.3-.7s-.1-.5-.3-.7zM5.8 14.7l6.2 6.3 6.2-6.3c.2-.2.3-.5.3-.7s-.1-.5-.3-.7c-.2-.2-.4-.3-.7-.3h-11c-.3 0-.5.1-.7.3-.2.2-.3.5-.3.7s.1.5.3.7z\"\/><\/svg><\/span><\/span><\/label><input type=\"checkbox\"  id=\"ez-toc-cssicon-toggle-item-6a2fcd8a261e0\" checked aria-label=\"Toggle\" \/><nav><ul class='ez-toc-list ez-toc-list-level-1 ' ><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-1\" href=\"https:\/\/buradabiliyorum.com\/en\/this-new-book-explores-the-difficulty-of-aligning-ai-with-our-values\/#Machine_learning_Mapping_inputs_to_outputs\" >Machine learning: Mapping inputs to outputs<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-2\" href=\"https:\/\/buradabiliyorum.com\/en\/this-new-book-explores-the-difficulty-of-aligning-ai-with-our-values\/#Reinforcement_learning_maximizing_rewards\" >Reinforcement learning: maximizing rewards<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-3\" href=\"https:\/\/buradabiliyorum.com\/en\/this-new-book-explores-the-difficulty-of-aligning-ai-with-our-values\/#Should_AI_imitate_humans\" >Should AI imitate humans?<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-4\" href=\"https:\/\/buradabiliyorum.com\/en\/this-new-book-explores-the-difficulty-of-aligning-ai-with-our-values\/#What_comes_next\" >What comes next?<\/a><\/li><\/ul><\/nav><\/div>\n<p>&#8220;<strong>#This new book explores the difficulty of aligning AI with our values<\/strong>&#8221;<\/p>\n<div>\n                                For decades, we\u2019ve been trying to develop artificial intelligence in our own image. And at every step of the way, we\u2019ve managed to create machines that can perform marvelous feats and at the same time make surprisingly dumb mistakes.<\/p>\n<p>After six decades of research and development, aligning AI systems with our goals, intents, and values continues to remain an elusive objective. Every major field of AI seems to solve part of the problem of replicating human intelligence while leaving out holes in critical areas. And these holes become problematic when we <a href=\"https:\/\/buradabiliyorum.com\/en\/category\/download-scripts-themes-apps\/\" data-internallinksmanager029f6b8e52c=\"9\" title=\"Download Scripts &amp; Themes &amp; Apps\" target=\"_blank\" rel=\"noopener\">app<\/a>ly current AI <a href=\"https:\/\/buradabiliyorum.com\/en\/category\/technology\/\" data-internallinksmanager029f6b8e52c=\"4\" title=\"Technology\" target=\"_blank\" rel=\"noopener\">technology<\/a> to areas where we expect intelligent agents to act with the rationality and logic we expect from humans.<\/p>\n<p>In his latest book,<span>\u00a0<\/span><em><a rel=\"nofollow noopener\" target=\"_blank\" href=\"https:\/\/brianchristian.org\/the-alignment-problem\/\">The Alignment Problem: Machine Learning and Human Values<\/a><\/em>, programmer and researcher Brian Christian discusses the challenges of making sure our AI models capture \u201cour norms and values, understand what we mean or intend, and, above all, do what we want.\u201d This is an issue that has become increasingly urgent in recent years, as<span>\u00a0<\/span><a rel=\"nofollow noopener\" target=\"_blank\" href=\"https:\/\/bdtechtalks.com\/2017\/08\/28\/artificial-intelligence-machine-learning-deep-learning\/\">machine learning<\/a>\u00a0has found its way into many fields and applications where making wrong decisions can have<span>\u00a0<\/span><a rel=\"nofollow noopener\" target=\"_blank\" href=\"https:\/\/bdtechtalks.com\/2020\/06\/10\/ai-weapons-of-math-destruction\/\">disastrous consequences<\/a>.<\/p>\n<p>As Christian describes: \u201cAs machine-learning systems grow not just increasingly pervasive but increasingly powerful, we will find ourselves more and more often in the position of the \u2018sorcerer\u2019s apprentice\u2019: we conjure a force, autonomous but totally compliant, give it a set of instructions, then scramble like mad to stop it once we realize our instructions are imprecise or incomplete\u2014lest we get, in some clever, horrible way, precisely what we asked for.\u201d<\/p>\n<p>In<span>\u00a0<\/span><em>The Alignment Problem,<span>\u00a0<\/span><\/em>Christian provides a thorough depiction of the current state of artificial intelligence and how we got here. He also discusses what\u2019s missing in different approaches to creating AI.<\/p>\n<p>Here are some key takeaways from the book.<\/p>\n<h2><span class=\"ez-toc-section\" id=\"Machine_learning_Mapping_inputs_to_outputs\"><\/span>Machine learning: Mapping inputs to outputs<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<div class=\"wp-block-image\">\n<figure class=\"alignleft size-large is-resized\"><a rel=\"nofollow noopener\" target=\"_blank\" href=\"https:\/\/i2.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2021\/01\/The-alignment-problem-book-cover.jpg?ssl=1\"><img loading=\"lazy\" decoding=\"async\" class=\"wp-image-9294 jetpack-lazy-image jetpack-lazy-image--handled lazy\" sizes=\"auto, (max-width: 331px) 100vw, 331px\" alt=\"The alignment problem book cover\" width=\"331\" height=\"512\" data-attachment-id=\"9294\" data-permalink=\"https:\/\/bdtechtalks.com\/2021\/01\/18\/ai-alignment-problem-brian-christian\/the-alignment-problem-book-cover\/\" data-orig-file=\"https:\/\/i2.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2021\/01\/The-alignment-problem-book-cover.jpg?fit=775%2C1200&amp;ssl=1\" data-orig-size=\"775,1200\" data-comments-opened=\"1\" data-image-meta=\"{\" aperture=\"\" data-image-title=\"The alignment problem book cover\" data-image-description=\"\" data-medium-file=\"https:\/\/i2.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2021\/01\/The-alignment-problem-book-cover.jpg?fit=194%2C300&amp;ssl=1\" data-large-file=\"https:\/\/i2.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2021\/01\/The-alignment-problem-book-cover.jpg?fit=661%2C1024&amp;ssl=1\" data-recalc-dims=\"1\" data-lazy-loaded=\"1\" src=\"https:\/\/i2.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2021\/01\/The-alignment-problem-book-cover.jpg?resize=331%2C512&amp;ssl=1\" data-lazy=\"true\" srcset=\"https:\/\/i2.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2021\/01\/The-alignment-problem-book-cover.jpg?resize=661%2C1024&amp;ssl=1 661w, https:\/\/i2.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2021\/01\/The-alignment-problem-book-cover.jpg?resize=194%2C300&amp;ssl=1 194w, https:\/\/i2.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2021\/01\/The-alignment-problem-book-cover.jpg?resize=768%2C1189&amp;ssl=1 768w, https:\/\/i2.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2021\/01\/The-alignment-problem-book-cover.jpg?resize=696%2C1078&amp;ssl=1 696w, https:\/\/i2.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2021\/01\/The-alignment-problem-book-cover.jpg?resize=271%2C420&amp;ssl=1 271w, https:\/\/i2.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2021\/01\/The-alignment-problem-book-cover.jpg?w=775&amp;ssl=1 775w\"\/><\/a><\/figure>\n<\/div>\n<p>In the earlier decades of AI research,<span>\u00a0<\/span><a rel=\"nofollow noopener\" target=\"_blank\" href=\"https:\/\/bdtechtalks.com\/2019\/11\/18\/what-is-symbolic-artificial-intelligence\/\">symbolic systems<\/a><span>\u00a0<\/span>made remarkable inroads in solving complicated problems that required logical reasoning. Yet they were terrible at simple tasks that every human learns at a young age, such as detecting objects, people, voices, and sounds. They also didn\u2019t scale well and required a lot of manual effort to create the rules and knowledge that defined their behavior.<\/p>\n<p>More recently, growing interest in machine learning and deep learning have helped advance<span>\u00a0<\/span><a rel=\"nofollow noopener\" target=\"_blank\" href=\"https:\/\/bdtechtalks.com\/2019\/01\/14\/what-is-computer-vision\/\">computer vision<\/a>, speech recognition, and<span>\u00a0<\/span><a rel=\"nofollow noopener\" target=\"_blank\" href=\"https:\/\/bdtechtalks.com\/2018\/02\/20\/ai-machine-learning-nlg-nlp\/\">natural language processing<\/a>, the very fields that symbolic AI struggled at. Machine learning algorithms\u00a0<a rel=\"nofollow noopener\" target=\"_blank\" href=\"https:\/\/bdtechtalks.com\/2019\/03\/25\/richard-sutton-artificial-intelligence-research\/\">scale well with the availability of data and compute resources<\/a>, which is largely why they\u2019ve become so popular in the past decade.<\/p>\n<p>But despite their remarkable achievements, machine learning algorithms are at their core complex mathematical functions that map observations to outcomes. Therefore, they\u2019re as good as their data and they start to break as the data they face in the world starts to deviate from examples they\u2019ve seen during training.<\/p>\n<p>In<span>\u00a0<\/span><em>The Alignment Problem<\/em>, Christian goes through many examples where machine learning algorithms have caused embarrassing and damaging failures. A popular example is a Google Photos classification algorithm that<span>\u00a0<\/span><a rel=\"nofollow noopener\" target=\"_blank\" href=\"https:\/\/www.forbes.com\/sites\/mzhang\/2015\/07\/01\/google-photos-tags-two-african-americans-as-gorillas-through-facial-recognition-software\/?sh=79e9c6cb713d\">tagged dark-skinned people as gorillas<\/a>. The problem was not with the AI algorithm but with the training data. Had Google trained the model on more examples of people with dark skin, it could have avoided the disaster.<\/p>\n<p>\u201cThe problem, of course, with a system that can, in theory, learn just about anything from a set of examples is that it finds itself, then, at the mercy of the examples from which it\u2019s taught,\u201d Christian writes.<\/p>\n<p>What\u2019s worse is that machine learning models can\u2019t tell right from wrong and make moral decisions. Whatever problem exists in a machine learning model\u2019s training data will be reflected in the model\u2019s behavior, often in nuanced and inconspicuous ways. For instance, in 2018, Amazon\u00a0<a rel=\"nofollow noopener\" target=\"_blank\" href=\"https:\/\/www.reuters.com\/article\/us-amazon-com-jobs-automation-insight\/amazon-scraps-secret-ai-recruiting-tool-that-showed-bias-against-women-idUSKCN1MK08G\">shut down a machine learning tool<\/a>\u00a0used in making hiring decisions because its decisions were biased against women. Obviously, none of the AI\u2019s creators wanted the model to select candidates based on their gender. In this case, the model, which was trained on the company\u2019s historical hiring data, reflected problems within Amazon itself.<\/p>\n<p>This is just one of the several cases where a machine learning model has picked up biases that existed in its training data and amplified them in its own unique ways. It is also a warning against trusting machine learning models that are trained on data we blindly collect from our own past behavior.<\/p>\n<p>\u201cModeling the world as it is is one thing. But as soon as you begin<span>\u00a0<\/span><em>using<\/em><span>\u00a0<\/span>that model, you are<span>\u00a0<\/span><em>changing<\/em><span>\u00a0<\/span>the world, in ways large and small. There is a broad assumption underlying many machine-learning models that the model itself will not<span>\u00a0<\/span><em>change<\/em><span>\u00a0<\/span>the reality it\u2019s modeling. In almost all cases, this is false,\u201d Christian writes. \u201cIndeed, uncareful deployment of these models might produce a feedback loop from which recovery becomes ever more difficult or requires ever greater interventions.\u201d<\/p>\n<p><em>[Read:\u00a0<span dir=\"auto\">How this company leveraged AI to become the Netflix of Finland<\/span>]<\/em><\/p>\n<p>Human intelligence has a lot to do with gathering data, finding patterns, and turning those patterns into actions. But while we usually try to simplify intelligent decision-making into a small set of inputs and outputs, the challenges of machine learning show that our assumptions about data and machine learning often turn out to be false.<\/p>\n<p>\u201cWe need to consider critically\u2026 not only where we get our training data but where we get the labels that will function in the system as a stand-in for ground truth. Often the ground truth is not the ground truth,\u201d Christian warns.<\/p>\n<h2><span class=\"ez-toc-section\" id=\"Reinforcement_learning_maximizing_rewards\"><\/span>Reinforcement learning: maximizing rewards<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<div class=\"wp-block-image\">\n<figure class=\"aligncenter size-large\"><a rel=\"nofollow noopener\" target=\"_blank\" href=\"https:\/\/i1.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2021\/01\/OpenAI-dota-2-reinforcement-learning.jpg?ssl=1\"><img loading=\"lazy\" decoding=\"async\" class=\"wp-image-9296 jetpack-lazy-image jetpack-lazy-image--handled lazy\" sizes=\"auto, (max-width: 696px) 100vw, 696px\" alt=\"OpenAI dota 2 reinforcement learning\" width=\"696\" height=\"371\" data-attachment-id=\"9296\" data-permalink=\"https:\/\/bdtechtalks.com\/2021\/01\/18\/ai-alignment-problem-brian-christian\/openai-dota-2-reinforcement-learning\/\" data-orig-file=\"https:\/\/i1.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2021\/01\/OpenAI-dota-2-reinforcement-learning.jpg?fit=1680%2C895&amp;ssl=1\" data-orig-size=\"1680,895\" data-comments-opened=\"1\" data-image-meta=\"{\" aperture=\"\" data-image-title=\"OpenAI dota 2 reinforcement learning\" data-image-description=\"\" data-medium-file=\"https:\/\/i1.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2021\/01\/OpenAI-dota-2-reinforcement-learning.jpg?fit=300%2C160&amp;ssl=1\" data-large-file=\"https:\/\/i1.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2021\/01\/OpenAI-dota-2-reinforcement-learning.jpg?fit=696%2C371&amp;ssl=1\" data-recalc-dims=\"1\" data-lazy-loaded=\"1\" src=\"https:\/\/i1.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2021\/01\/OpenAI-dota-2-reinforcement-learning.jpg?resize=696%2C371&amp;ssl=1\" data-lazy=\"true\" srcset=\"https:\/\/i1.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2021\/01\/OpenAI-dota-2-reinforcement-learning.jpg?resize=1024%2C546&amp;ssl=1 1024w, https:\/\/i1.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2021\/01\/OpenAI-dota-2-reinforcement-learning.jpg?resize=300%2C160&amp;ssl=1 300w, https:\/\/i1.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2021\/01\/OpenAI-dota-2-reinforcement-learning.jpg?resize=768%2C409&amp;ssl=1 768w, https:\/\/i1.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2021\/01\/OpenAI-dota-2-reinforcement-learning.jpg?resize=1536%2C818&amp;ssl=1 1536w, https:\/\/i1.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2021\/01\/OpenAI-dota-2-reinforcement-learning.jpg?resize=696%2C371&amp;ssl=1 696w, https:\/\/i1.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2021\/01\/OpenAI-dota-2-reinforcement-learning.jpg?resize=1068%2C569&amp;ssl=1 1068w, https:\/\/i1.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2021\/01\/OpenAI-dota-2-reinforcement-learning.jpg?resize=788%2C420&amp;ssl=1 788w, https:\/\/i1.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2021\/01\/OpenAI-dota-2-reinforcement-learning.jpg?w=1680&amp;ssl=1 1680w, https:\/\/i1.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2021\/01\/OpenAI-dota-2-reinforcement-learning.jpg?w=1392&amp;ssl=1 1392w\"\/><\/a><figcaption><span id=\"urn:enhancement-200\" class=\"textannotation\"><br \/>Reinforcement learning<\/span> has helped researchers create AI that achieves remarkable feats such as beating champions at complicated video <a href=\"https:\/\/buradabiliyorum.com\/en\/category\/game\/\" data-internallinksmanager029f6b8e52c=\"7\" title=\"Game\" target=\"_blank\" rel=\"noopener\">game<\/a>s.<\/figcaption><\/figure>\n<\/div>\n<p>Another branch of AI that has gained much traction in the past decade is<span>\u00a0<\/span><a rel=\"nofollow noopener\" target=\"_blank\" href=\"https:\/\/bdtechtalks.com\/2019\/05\/28\/what-is-reinforcement-learning\/\">reinforcement learning<\/a>, a subset of machine learning in which the model is given the rules of a problem space and a reward function. The model is then left to explore the space for itself and find ways to maximize its rewards.<\/p>\n<p>\u201cReinforcement learning\u2026 offers us a powerful, and perhaps even universal, definition of what intelligence<span>\u00a0<\/span><em>is<\/em>,\u201d Christian writes. \u201cIf intelligence is, as computer scientist John McCarthy famously said, \u2018the computational part of the ability to achieve goals in the world,\u2019 then reinforcement learning offers a strikingly <a href=\"https:\/\/buradabiliyorum.com\/en\/category\/general\/\" data-internallinksmanager029f6b8e52c=\"3\" title=\"General\" target=\"_blank\" rel=\"noopener\">general<\/a> toolbox for doing so. Indeed it is likely that its core principles were stumbled into by evolution time and again\u2014and it is likely that they will form the bedrock of whatever artificial intelligence the twenty-first century has in store.\u201d<\/p>\n<p>Reinforcement learning is behind great scientific achievements such as AI systems that have mastered Atari games, Go, StarCraft 2, and DOTA 2. It has also found many uses in robotics. But each of those achievements also proves that purely pursuing external rewards is not exactly how intelligence works.<\/p>\n<p>For one thing, reinforcement learning models require massive amounts of training cycles to obtain simple results. For this very reason, research in this field has been limited to a few labs that are backed by<span>\u00a0<\/span><a rel=\"nofollow noopener\" target=\"_blank\" href=\"https:\/\/bdtechtalks.com\/2020\/08\/17\/openai-gpt-3-commercial-ai\/\">very wealthy companies<\/a>. Reinforcement learning systems are also very rigid. For instance, a reinforcement learning model that plays StarCraft 2 at championship level won\u2019t be able to play another game with similar mechanics. Reinforcement learning agents also tend to get stuck in meaningless loops that maximize a simple reward at the expense of long-term goals. An example is this boat-racing AI that has managed to hack its environment by continuously collecting bonus items without considering the greater goal of winning the race.<\/p>\n<figure class=\"wp-block-embed is-type-video is-provider-youtube wp-block-embed-youtube wp-embed-aspect-4-3 wp-has-aspect-ratio\"><iframe loading=\"lazy\" title=\"CoastRunners 7\" width=\"640\" height=\"480\" src=\"https:\/\/www.youtube.com\/embed\/tlOIHko8ySg?feature=oembed\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share\" referrerpolicy=\"strict-origin-when-cross-origin\" allowfullscreen><\/iframe><br \/>\n<\/figure>\n<p>\u201cUnplugging the hardwired external rewards may be a necessary part of building truly general AI: because life, unlike an Atari game, emphatically does not come pre-labeled with <span style=\"background-color: rgba(46, 146, 255, 0.2);\">real-time<\/span>\u00a0feedback on how good or bad each of our actions is,\u201d Christian writes. \u201cWe have parents and teachers, sure, who can correct our spelling and pronunciation and, occasionally, our behavior. But this hardly covers a fraction of what we do and say and think, and the authorities in our life do not always agree. Moreover, it is one of the central rites of passage of the human condition that we must learn to make these judgments by our own lights and for ourselves.\u201d<\/p>\n<p>Christian also suggests that while reinforcement learning starts with rewards and develops behavior that maximizes those rewards, the reverse is perhaps even more interesting and critical: \u201cGiven the behavior, we want from our machines, how do we structure the environment\u2019s rewards to bring that behavior about? How do we get what we want when it is<span>\u00a0<\/span><em>we<\/em><span>\u00a0<\/span>who sit in the back of the audience, in the critic\u2019s chair\u2014<em>we<\/em><span>\u00a0<\/span>who administer the food pellets, or their digital equivalent?\u201d<\/p>\n<h2><span class=\"ez-toc-section\" id=\"Should_AI_imitate_humans\"><\/span>Should AI imitate humans?<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<div class=\"wp-block-image\">\n<figure class=\"aligncenter size-large\"><a rel=\"nofollow noopener\" target=\"_blank\" href=\"https:\/\/i0.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2019\/07\/machine-learning-artificial-intelligence.jpg?ssl=1\"><img loading=\"lazy\" decoding=\"async\" class=\"wp-image-5089 jetpack-lazy-image jetpack-lazy-image--handled lazy\" sizes=\"auto, (max-width: 696px) 100vw, 696px\" alt=\"machine learning artificial intelligence\" width=\"696\" height=\"359\" data-attachment-id=\"5089\" data-permalink=\"https:\/\/bdtechtalks.com\/2019\/07\/08\/ai-bias-survival-of-the-best-fit\/machine-learning-artificial-intelligence\/\" data-orig-file=\"https:\/\/i0.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2019\/07\/machine-learning-artificial-intelligence.jpg?fit=4109%2C2117&amp;ssl=1\" data-orig-size=\"4109,2117\" data-comments-opened=\"1\" data-image-meta=\"{\" aperture=\"\" data-image-title=\"machine learning artificial intelligence\" data-image-description=\"\" data-medium-file=\"https:\/\/i0.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2019\/07\/machine-learning-artificial-intelligence.jpg?fit=300%2C155&amp;ssl=1\" data-large-file=\"https:\/\/i0.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2019\/07\/machine-learning-artificial-intelligence.jpg?fit=696%2C359&amp;ssl=1\" data-recalc-dims=\"1\" data-lazy-loaded=\"1\" src=\"https:\/\/i0.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2019\/07\/machine-learning-artificial-intelligence.jpg?resize=696%2C359&amp;ssl=1\" data-lazy=\"true\" srcset=\"https:\/\/i0.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2019\/07\/machine-learning-artificial-intelligence.jpg?resize=1024%2C528&amp;ssl=1 1024w, https:\/\/i0.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2019\/07\/machine-learning-artificial-intelligence.jpg?resize=300%2C155&amp;ssl=1 300w, https:\/\/i0.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2019\/07\/machine-learning-artificial-intelligence.jpg?resize=768%2C396&amp;ssl=1 768w, https:\/\/i0.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2019\/07\/machine-learning-artificial-intelligence.jpg?resize=696%2C359&amp;ssl=1 696w, https:\/\/i0.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2019\/07\/machine-learning-artificial-intelligence.jpg?resize=1068%2C550&amp;ssl=1 1068w, https:\/\/i0.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2019\/07\/machine-learning-artificial-intelligence.jpg?resize=815%2C420&amp;ssl=1 815w, https:\/\/i0.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2019\/07\/machine-learning-artificial-intelligence.jpg?resize=1920%2C989&amp;ssl=1 1920w, https:\/\/i0.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2019\/07\/machine-learning-artificial-intelligence.jpg?w=1392&amp;ssl=1 1392w, https:\/\/i0.wp.com\/bdtechtalks.com\/wp-content\/uploads\/2019\/07\/machine-learning-artificial-intelligence.jpg?w=2088&amp;ssl=1 2088w\"\/><\/a><\/figure>\n<\/div>\n<p>In<span>\u00a0<\/span><em>The Alignment Problem<\/em>, Christian also discusses the implications of developing AI agents that learn through pure imitation of human actions. An example is self-driving cars that learn by observing how humans drive.<\/p>\n<p>Imitation can do wonders, especially in problems where the rules and labels are not clear-cut. But again, imitation paints an incomplete picture of the intelligence puzzle. We humans learn a lot through imitation and rote learning, especially at a young age. But imitation is but one of several mechanisms we use to develop intelligent behavior. As we observe the behavior of others, we also adapt our own version of that behavior that is aligned with our own limits, intents, goals, needs, and values.<\/p>\n<p>\u201cIf someone is fundamentally faster or stronger or differently sized than you, or quicker-thinking than you could ever be, mimicking their actions to perfection may still not work,\u201d Christian writes. \u201cIndeed, it may be catastrophic. You\u2019ll do what you<span>\u00a0<\/span><em>would<\/em><span>\u00a0do<\/span> if you were them. But you\u2019re not them. And what you do is not what<span>\u00a0<\/span><em>they<\/em><span>\u00a0<\/span>would do if they were<span>\u00a0<\/span><em>you<\/em>.\u201d<\/p>\n<p>In other cases, AI systems use imitation to observe and predict our behavior and try to assist us. But this too presents a challenge. AI systems are not bound by the same constraints and limits as we are, and they often misinterpret our intentions and what\u2019s good for us. Instead of protecting us against our bad habits,<span>\u00a0<\/span><a rel=\"nofollow noopener\" target=\"_blank\" href=\"https:\/\/bdtechtalks.com\/2019\/05\/20\/artificial-intelligence-filter-bubbles-news-bias\/\">they amplify them<\/a><span>\u00a0<\/span>and they push us toward acquiring the bad habits of others. And they\u2019re becoming pervasive in every aspect of our lives.<\/p>\n<p>\u201cOur digital butlers are watching closely,\u201d Christian writes. \u201cThey see our private as well as our public lives, our best and worst selves, without necessarily knowing which is which or making a distinction at all. They, by and large, reside in a kind of uncanny valley of sophistication: able to infer sophisticated models of our desires from our behavior, but unable to be taught, and disinclined to cooperate. They\u2019re thinking hard about what we are going to do next, about how they might make their next commission, but they don\u2019t seem to understand what we want, much less who we hope to become.\u201d<\/p>\n<h2><span class=\"ez-toc-section\" id=\"What_comes_next\"><\/span>What comes next?<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>Advances in machine learning show how far we\u2019ve come toward the goal of creating thinking machines. But the challenges of machine learning and the alignment problem also remind us of how much more we have to learn before we can create<span>\u00a0<\/span><a rel=\"nofollow noopener\" target=\"_blank\" href=\"https:\/\/bdtechtalks.com\/2020\/05\/13\/what-is-artificial-general-intelligence-agi\/\">human-level intelligence<\/a>.<\/p>\n<p>AI scientists and researchers are exploring<span>\u00a0<\/span><a rel=\"nofollow noopener\" target=\"_blank\" href=\"https:\/\/bdtechtalks.com\/2019\/12\/23\/yoshua-bengio-neurips-2019-deep-learning\/\">several<\/a><span>\u00a0<\/span><a rel=\"nofollow noopener\" target=\"_blank\" href=\"https:\/\/bdtechtalks.com\/2020\/03\/04\/gary-marcus-hybrid-ai\/\">different<\/a><span>\u00a0<\/span><a rel=\"nofollow noopener\" target=\"_blank\" href=\"https:\/\/bdtechtalks.com\/2020\/03\/23\/yann-lecun-self-supervised-learning\/\">ways<\/a><span>\u00a0<\/span>to overcome these hurdles and create AI systems that can benefit humanity without causing harm. Until then, we\u2019ll have to tread carefully and beware of how much credit we assign to systems that mimic human intelligence on the surface.<\/p>\n<p>\u201cOne of the most dangerous things one can do in machine learning\u2014and otherwise\u2014is to find a model that is reasonably good, declare victory, and henceforth begin to confuse the map with the territory,\u201d Christian warns<em>.<\/em><\/p>\n<p><i><span style=\"font-weight: 400;\">This article was originally published by Ben Dickson on <\/span><\/i><a rel=\"nofollow noopener\" target=\"_blank\" href=\"https:\/\/bdtechtalks.com\/\"><i><span style=\"font-weight: 400;\">TechTalks<\/span><\/i><\/a><i><span style=\"font-weight: 400;\">, a publication that examines trends in technology, how they affect the way we live and do business, and the problems they solve. But we also discuss the evil side of technology, the darker implications of new tech and what we need to look out for. You can read the original article <a rel=\"nofollow noopener\" target=\"_blank\" href=\"https:\/\/bdtechtalks.com\/2021\/01\/18\/ai-alignment-problem-brian-christian\/\">here<\/a>.<\/span><\/i><\/p>\n<p class=\"c-post-pubDate\">\n                                    Published January 30, 2021 \u2014 15:00 UTC\n                                <\/p>\n<\/p><\/div>\n<p><script data-src=\"https:\/\/connect.facebook.net\/en_US\/sdk.js#xfbml=1&amp;appId=378011798897423&amp;version=v2.6\" id=\"socialSrcFacebook\" type=\"text\/template\"><\/script><\/p>\n<blockquote><p><strong><span style=\"color: #ff6600;\">If you liked the article, do not forget to share it with your friends. Follow us on\u00a0<span style=\"color: #ff0000;\"><a style=\"color: #ff0000;\" href=\"https:\/\/news.google.com\/publications\/CAAqBwgKMLG0nwswvr63Aw\" target=\"_blank\" rel=\"nofollow noopener noreferrer\">Google News<\/a><\/span>\u00a0too, click on the star and choose us from your favorites.<\/span><\/strong><\/p><\/blockquote>\n<blockquote>\n<p style=\"text-align: center;\">For forums sites go to <span style=\"color: #ff9900;\"><a style=\"color: #ff9900;\" href=\"https:\/\/forum.buradabiliyorum.com\/\" target=\"_blank\" rel=\"noopener\">Forum.BuradaBiliyorum.Com<\/a><\/span><\/strong><\/p>\n<\/blockquote>\n<blockquote>\n<p style=\"text-align: center;\"><strong>If you want to read more like this article, you can visit our <span style=\"color: #ff9900;\"><a style=\"color: #ff9900;\" href=\"https:\/\/en.buradabiliyorum.com\/technology\/\" target=\"_blank\" rel=\"noopener\">Technology category.<\/a><\/span><\/strong><\/p>\n<\/blockquote>\n<p><span style=\"color: black;\"><a style=\"color: #ff9900;\" href=\"https:\/\/thenextweb.com\/neural\/2021\/01\/30\/this-new-book-explores-the-difficulty-of-aligning-ai-with-our-values-syndication\/\" target=\"_blank\" rel=\"noopener\">Source<\/a><\/span><\/p>\n","protected":false},"excerpt":{"rendered":"<p>&#8220;#This new book explores the difficulty of aligning AI with our values&#8221; For decades, we\u2019ve been trying to develop artificial intelligence in our own image. And at every step of the way, we\u2019ve managed to create machines that can perform marvelous feats and at the same time make surprisingly dumb mistakes. After six decades of&#8230;<\/p>\n","protected":false},"author":1,"featured_media":166868,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"fifu_image_url":"https:\/\/img-cdn.tnwcdn.com\/image\/neural?filter_last=1&fit=1280,640&url=https:\/\/cdn0.tnwcdn.com\/wp-content\/blogs.dir\/1\/files\/2021\/01\/1-copy-62.jpg&signature=0f9ea9779549fc5b008bbac649d3db98","fifu_image_alt":"","footnotes":""},"categories":[18],"tags":[],"class_list":["post-166867","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-technology"],"_links":{"self":[{"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/posts\/166867","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/comments?post=166867"}],"version-history":[{"count":0,"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/posts\/166867\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/media\/166868"}],"wp:attachment":[{"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/media?parent=166867"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/categories?post=166867"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/tags?post=166867"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}