{"id":48573,"date":"2026-09-07T11:54:12","date_gmt":"2026-09-07T09:54:12","guid":{"rendered":"https:\/\/media-and-learning.eu\/?p=48573"},"modified":"2026-09-07T11:57:42","modified_gmt":"2026-09-07T09:57:42","slug":"misconceptions-about-artificial-intelligence-what-we-think-we-know-and-what-we-actually-need-to-learn","status":"publish","type":"post","link":"https:\/\/media-and-learning.eu\/subject\/artificial-intelligence\/misconceptions-about-artificial-intelligence-what-we-think-we-know-and-what-we-actually-need-to-learn\/","title":{"rendered":"(Mis)conceptions about Artificial Intelligence: what we think we know\u2014and what we actually need to learn"},"content":{"rendered":"\n<p class=\"wp-block-paragraph\">by <strong>Tobias Bahr<\/strong>, Heidelberg University, Germany.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Students, according to constructivism (Ben-Ari, 2001) make sense, build <em>conceptions<\/em> about AI by engaging with AI tools in their everyday life, e.g. a social media app with a personalised feed or advertisement, using generative AI (GenAI) to answer questions, guidance during moments of anxiety or creating images.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">As most AI systems are by design opaque and word like \u201clearning\u201d, \u201cintelligence\u201d and \u201cautonomous agents\u201d are used in everyday conversations and are present in media, initial conceptions that do not match scientific definitions (Bahr, 2026) are built.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Why do some conceptions persist?<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">When we encounter something as abstract as machine learning, we reach for familiar analogies &#8211; often anthropomorphic ones. AI &#8220;learns,&#8221; so it must learn like us. AI &#8220;decides,&#8221; so it must have intentions.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Research in <em>science<\/em> education made three characteristics of conceptions (Niedderer &amp; Schecker, 1992):<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Resilient &#8211; they don&#8217;t disappear after a single lesson (or workshop).<\/li>\n\n\n\n<li>Flexible &#8211; the same person may hold contradictory beliefs in different contexts and projects conceptions onto other conceptions.<\/li>\n\n\n\n<li>Potentially limited &#8211; there may only be a handful of core conceptions.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">However, for computer science education, especially in AI, new conceptions arise with the sheer amount of new empirical studies. Less is known about the structure of conceptions. However, some conceptions seem to be resilient (or stable) even after an intervention (cf. Bahr, 2026).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Addressing conceptions and shifting them to more sophisticated one\u2019s is not about delivering facts but creating learning activities for <em>all<\/em> students that are perceived as relevant to their everyday life, focus on problem-based empirically tested approaches, and including the impact of AI (cf. European Commission, 2026).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>How to assess them?<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Several approaches have been used to assess students\u2019, teachers\u2019, and non-experts\u2019 conceptions about AI. The most research-economic one is a concept inventory. A mostly multiple choice questionnaire with answers derived from interview data. If you are interested and want to participate in a study about non-experts conceptions, as well as reflect upon your own conceptions you can click on<a href=\"https:\/\/limesurvey.urz.uni-heidelberg.de\/index.php\/123665?lang=en\" target=\"_blank\" rel=\"noreferrer noopener\"> this link<\/a>.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">You can check your answers in a solution sheet at the end. The goal with this questionnaire to give teachers practical tools for diagnosing what their students believe before instruction, so they can tailor their teaching accordingly.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>What are maybe important conceptions to know?<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Here are a few of the most persistent misunderstandings my colleagues and I have documented (Bahr, 2026; Mannila et al., 2025):<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>&#8220;More data is always better.&#8221; Many people assume that feeding an AI every available data point will automatically improve its performance. However, data quality, relevance, and diversity often matter far more than sheer volume. Irrelevant data can actively lower the performance of a model.<\/li>\n\n\n\n<li>&#8220;AI can generalise like humans.&#8221; A model trained to recognise cats in photographs cannot suddenly identify dogs &#8211; or even cats in unusual lighting &#8211; without additional training. Transfer learning exists, but it&#8217;s not magic. This same goes for current flagship GenAI models. They are just stochastic parrots with good marketing. If an AI chatbot says its hungry it just repeats the given data that is accruing the most.<\/li>\n\n\n\n<li>&#8220;Programmers understand exactly how AI reaches its conclusions.&#8221; Even experts often cannot fully explain why a neural network made a particular decision. This has profound implications for accountability and trust.<\/li>\n\n\n\n<li>&#8220;A strong AI already exists somewhere.&#8221; General AI, a system capable of any intellectual task a human can perform, remains science fiction. Every AI you interact with today is &#8220;narrow,&#8221; specialised for a specific task (e.g. text generation).<\/li>\n\n\n\n<li>&#8220;AI decisions are objective.&#8221; Because AI relies on data created by humans, it can encode and amplify human biases. An algorithm is only as fair as the training data and design choices behind it.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>A path forward<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">When learners articulate their existing beliefs, through concept cartoons, open-ended questions, or simple discussions, educators can meet them where they are. From there, carefully designed activities can create the cognitive dissonance necessary for genuine learning (cf. Bahr, 2026).<\/p>\n\n\n\n<h3 class=\"wp-block-heading\"><a href=\"https:\/\/media-and-learning.eu\/files\/2026\/09\/Refereces.pdf\" target=\"_blank\" rel=\"noreferrer noopener\">References<\/a><\/h3>\n\n\n\n<div class=\"wp-block-media-text is-stacked-on-mobile\" style=\"grid-template-columns:22% auto\"><figure class=\"wp-block-media-text__media\"><img loading=\"lazy\" decoding=\"async\" width=\"811\" height=\"1024\" src=\"https:\/\/media-and-learning.eu\/files\/2026\/09\/Bild_Bahr-811x1024.jpg\" alt=\"\" class=\"wp-image-48575 size-full\" srcset=\"https:\/\/media-and-learning.eu\/files\/2026\/09\/Bild_Bahr-811x1024.jpg 811w, https:\/\/media-and-learning.eu\/files\/2026\/09\/Bild_Bahr-238x300.jpg 238w, https:\/\/media-and-learning.eu\/files\/2026\/09\/Bild_Bahr-768x970.jpg 768w, https:\/\/media-and-learning.eu\/files\/2026\/09\/Bild_Bahr-1216x1536.jpg 1216w, https:\/\/media-and-learning.eu\/files\/2026\/09\/Bild_Bahr-370x467.jpg 370w, https:\/\/media-and-learning.eu\/files\/2026\/09\/Bild_Bahr-270x341.jpg 270w, https:\/\/media-and-learning.eu\/files\/2026\/09\/Bild_Bahr-570x720.jpg 570w, https:\/\/media-and-learning.eu\/files\/2026\/09\/Bild_Bahr-740x935.jpg 740w, https:\/\/media-and-learning.eu\/files\/2026\/09\/Bild_Bahr-600x758.jpg 600w, https:\/\/media-and-learning.eu\/files\/2026\/09\/Bild_Bahr.jpg 1285w\" sizes=\"auto, (max-width: 811px) 100vw, 811px\" \/><\/figure><div class=\"wp-block-media-text__content\">\n<p class=\"wp-block-paragraph\"><strong>Dr. phil. Tobias Bahr<\/strong> is a postdoc researcher at Heidelberg University addressing AI in teacher education at the Heidelberg School of Education.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">His main research expertise lies within qualitative and quantitative methods and empirical research in interdisciplinary STEM and especially K-12 computer science education.<\/p>\n<\/div><\/div>\n","protected":false},"excerpt":{"rendered":"<p>by Tobias Bahr, Heidelberg University, Germany. Students, according to constructivism (Ben-Ari, 2001) make sense, build conceptions about AI by engaging with AI tools in their everyday life, e.g. a social media app with a personalised feed or advertisement, using generative AI (GenAI) to answer questions, guidance during moments of anxiety or creating images. As most [&hellip;]<\/p>\n","protected":false},"author":7,"featured_media":48577,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_mo_disable_npp":"","footnotes":""},"categories":[362,4,275],"tags":[],"class_list":["post-48573","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-artificial-intelligence","category-featured-articles","category-higher-education"],"featured_image_src":"https:\/\/media-and-learning.eu\/files\/2026\/09\/Website-Banner-700-x-200-px-1000-x-300-px-85.png","author_info":{"display_name":"Shirin Izadpanah","author_link":"https:\/\/media-and-learning.eu\/author\/shirin-izadpanah\/"},"_links":{"self":[{"href":"https:\/\/media-and-learning.eu\/api-json\/wp\/v2\/posts\/48573","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/media-and-learning.eu\/api-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/media-and-learning.eu\/api-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/media-and-learning.eu\/api-json\/wp\/v2\/users\/7"}],"replies":[{"embeddable":true,"href":"https:\/\/media-and-learning.eu\/api-json\/wp\/v2\/comments?post=48573"}],"version-history":[{"count":5,"href":"https:\/\/media-and-learning.eu\/api-json\/wp\/v2\/posts\/48573\/revisions"}],"predecessor-version":[{"id":48590,"href":"https:\/\/media-and-learning.eu\/api-json\/wp\/v2\/posts\/48573\/revisions\/48590"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/media-and-learning.eu\/api-json\/wp\/v2\/media\/48577"}],"wp:attachment":[{"href":"https:\/\/media-and-learning.eu\/api-json\/wp\/v2\/media?parent=48573"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/media-and-learning.eu\/api-json\/wp\/v2\/categories?post=48573"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/media-and-learning.eu\/api-json\/wp\/v2\/tags?post=48573"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}