{"id":34823,"date":"2024-08-20T11:34:19","date_gmt":"2024-08-20T09:34:19","guid":{"rendered":"https:\/\/www.l3s.de\/?p=34823"},"modified":"2024-09-05T12:16:56","modified_gmt":"2024-09-05T10:16:56","slug":"investigating-error-types-in-gpt-4","status":"publish","type":"post","link":"https:\/\/www.l3s.de\/de\/investigating-error-types-in-gpt-4\/","title":{"rendered":"Fehlertypen in GPT-4 untersuchen"},"content":{"rendered":"<p class=\"wp-block-paragraph\"><strong>L3S Bestes Paper des Quartals (Q2\/2024)&nbsp;<\/strong>&nbsp;<br><strong>Category: IR, Generative AI<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\" style=\"font-size:23px\"><strong>Jenseits der Genauigkeit: Untersuchung von Fehlertypen in GPT-4-Antworten auf USMLE-Fragen&nbsp;<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Autoren: Soumyadeep Roy, Aparup Khatua, Fatemeh Ghoochani, Uwe Hadler, Wolfgang Nejdl, Niloy Ganguly&nbsp;<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><em><br>Ver\u00f6ffentlicht auf der A*-Konferenz \"47th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR 2024)\u201d: <a href=\"https:\/\/dl.acm.org\/doi\/10.1145\/3626772.3657882\" title=\"\">https:\/\/dl.acm.org\/doi\/10.1145\/3626772.3657882<\/a><\/em><\/p>\n\n\n\n<div style=\"height:28px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Die Ver\u00f6ffentlichung in K\u00fcrze:<\/strong>&nbsp;<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Die Studie untersucht die Fehler von GPT-4, einem f\u00fchrenden KI-Modell, bei der Beantwortung komplexer medizinischer Fragen der United States Medical Licensing Examination (USMLE). Der entscheidende Beitrag dieser Arbeit ist die Feststellung, dass eine Begr\u00fcndung (Modellerkl\u00e4rungen in nat\u00fcrlicher Sprache) zusammen mit der Antwort durch das KI-Modell generiert werden muss, um KI-Modelle zur Beantwortung medizinischer Fragen zu evaluieren. Die Forscher haben mit Hilfe von medizinischen Fachleuten eine detaillierte Fehlertaxonomie entwickelt. Unsere gro\u00df angelegte Annotationsstudie umfasst 44 medizinische Experten, die GPT-4-Antworten auf Satzebene gem\u00e4\u00df der entwickelten Fehlertaxonomie annotierten. Die Forscher stellen fest, dass GPT-4 nur selten sachliche Fehler machen. Sie sind jedoch sehr anf\u00e4llig f\u00fcr Argumentationsfehler, die f\u00fcr Menschen und bestehende Moderationssysteme noch schwieriger zu erkennen sind. Diese Forschung liefert somit wertvolle Einblicke in die St\u00e4rken und Grenzen von GPT-4 bei der Beantwortung von medizinischen Fragen.&nbsp;<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Welches Problem l\u00f6sen Sie mit Ihrer Forschung?<\/strong>&nbsp;<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Diese Forschung befasst sich mit der Herausforderung zu verstehen, warum und wie fortschrittliche KI-Modelle wie GPT-4 bei der Beantwortung komplexer medizinischer Fragen Fehler machen. Durch die Entwicklung einer detaillierten Fehlertaxonomie und die Analyse der Antworten von GPT-4 liefert die Studie ein tieferes Verst\u00e4ndnis f\u00fcr den Entscheidungsprozess des Modells und seine Grenzen im medizinischen Bereich.&nbsp;<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Welche potenziellen Auswirkungen haben Ihre Ergebnisse?<\/strong>&nbsp;<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Die Ergebnisse dieser Studie k\u00f6nnen dazu beitragen, KI-Modelle f\u00fcr medizinische Anwendungen zu verbessern, indem sie die spezifischen Bereiche identifizieren, in denen sie Schwierigkeiten haben. Dies k\u00f6nnte zu zuverl\u00e4ssigeren KI-gest\u00fctzten medizinischen Entscheidungshilfen f\u00fchren, was die Patientenversorgung und die medizinische Ausbildung verbessern k\u00f6nnte. Dar\u00fcber hinaus liefert die Forschung wertvolle Ressourcen f\u00fcr weitere Studien zur KI-Leistung bei komplexen medizinischen Aufgaben.&nbsp;<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Was ist neu an Ihrer Forschung?<\/strong>&nbsp;<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Diese Studie stellt eine neuartige, dom\u00e4nenspezifische Fehlertaxonomie f\u00fcr KI-Antworten auf medizinische Fragen vor, die in Zusammenarbeit mit Medizinern entwickelt wurde.Au\u00dferdem wird ein neuer Datensatz mit den detaillierten Antworten von GPT-4 auf USMLE-Fragen vorgestellt, einschlie\u00dflich Erkl\u00e4rungen f\u00fcr die Auswahl der Antworten.Dieser Ansatz erm\u00f6glicht ein umfassenderes Verst\u00e4ndnis des Denkprozesses der KI \u00fcber einfache Genauigkeitsmessungen hinaus.&nbsp;<\/p>","protected":false},"excerpt":{"rendered":"<p>Die Untersuchung der Fehlertypen in GPT-4 wurde als beste Ver\u00f6ffentlichung des L3S in der Kategorie \"IR und Generative KI\" ausgezeichnet. <\/p>","protected":false},"author":11,"featured_media":34824,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[161],"tags":[],"class_list":["post-34823","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-best-publications"],"acf":[],"aioseo_notices":[],"aioseo_head":"\n\t\t<!-- All in One SEO 4.9.8 - aioseo.com -->\n\t<meta name=\"description\" content=\"The investigation of error types in GPT-4 is the best publication in the IR and Generative AI category.\" \/>\n\t<meta name=\"robots\" content=\"max-image-preview:large\" \/>\n\t<meta name=\"author\" content=\"borovynska\"\/>\n\t<link rel=\"canonical\" href=\"https:\/\/www.l3s.de\/de\/investigating-error-types-in-gpt-4\/\" \/>\n\t<meta name=\"generator\" content=\"All in One SEO (AIOSEO) 4.9.8\" \/>\n\t\t<meta property=\"og:locale\" content=\"de_DE\" \/>\n\t\t<meta property=\"og:site_name\" content=\"L3S Research Center - Trustworthy AI &amp; Digital Transformation\" \/>\n\t\t<meta property=\"og:type\" content=\"article\" \/>\n\t\t<meta property=\"og:title\" content=\"Investigating Error Types in GPT-4 - L3S Research Center\" \/>\n\t\t<meta property=\"og:description\" content=\"The investigation of error types in GPT-4 is the best publication in the IR and Generative AI category.\" \/>\n\t\t<meta property=\"og:url\" content=\"https:\/\/www.l3s.de\/de\/investigating-error-types-in-gpt-4\/\" \/>\n\t\t<meta property=\"og:image\" content=\"https:\/\/www.l3s.de\/wp-content\/uploads\/2024\/11\/LS3-Logo-2024-WHITE.png\" \/>\n\t\t<meta property=\"og:image:secure_url\" content=\"https:\/\/www.l3s.de\/wp-content\/uploads\/2024\/11\/LS3-Logo-2024-WHITE.png\" \/>\n\t\t<meta property=\"article:published_time\" content=\"2024-08-20T09:34:19+00:00\" \/>\n\t\t<meta property=\"article:modified_time\" content=\"2024-09-05T10:16:56+00:00\" \/>\n\t\t<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n\t\t<meta name=\"twitter:site\" content=\"@l3s_luh\" \/>\n\t\t<meta name=\"twitter:title\" content=\"Investigating Error Types in GPT-4 - L3S Research Center\" \/>\n\t\t<meta name=\"twitter:description\" content=\"The investigation of error types in GPT-4 is the best publication in the IR and Generative AI category.\" \/>\n\t\t<meta name=\"twitter:creator\" content=\"@l3s_luh\" \/>\n\t\t<meta name=\"twitter:image\" content=\"https:\/\/www.l3s.de\/wp-content\/uploads\/2024\/11\/LS3-Logo-2024-WHITE.png\" \/>\n\t\t<script type=\"application\/ld+json\" class=\"aioseo-schema\">\n\t\t\t{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"BlogPosting\",\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/investigating-error-types-in-gpt-4\\\/#blogposting\",\"name\":\"Investigating Error Types in GPT-4 - L3S Research Center\",\"headline\":\"Investigating Error Types in GPT-4\",\"author\":{\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/author\\\/borovynska\\\/#author\"},\"publisher\":{\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/#organization\"},\"image\":{\"@type\":\"ImageObject\",\"url\":\"https:\\\/\\\/www.l3s.de\\\/wp-content\\\/uploads\\\/2024\\\/08\\\/IR-Generative-AI.png\",\"width\":1200,\"height\":627},\"datePublished\":\"2024-08-20T11:34:19+02:00\",\"dateModified\":\"2024-09-05T12:16:56+02:00\",\"inLanguage\":\"de-DE\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/investigating-error-types-in-gpt-4\\\/#webpage\"},\"isPartOf\":{\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/investigating-error-types-in-gpt-4\\\/#webpage\"},\"articleSection\":\"Best Publications\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/investigating-error-types-in-gpt-4\\\/#breadcrumblist\",\"itemListElement\":[{\"@type\":\"ListItem\",\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de#listItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/www.l3s.de\\\/de\",\"nextItem\":{\"@type\":\"ListItem\",\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/category\\\/best-publications\\\/#listItem\",\"name\":\"Best Publications\"}},{\"@type\":\"ListItem\",\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/category\\\/best-publications\\\/#listItem\",\"position\":2,\"name\":\"Best Publications\",\"item\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/category\\\/best-publications\\\/\",\"nextItem\":{\"@type\":\"ListItem\",\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/investigating-error-types-in-gpt-4\\\/#listItem\",\"name\":\"Investigating Error Types in GPT-4\"},\"previousItem\":{\"@type\":\"ListItem\",\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de#listItem\",\"name\":\"Home\"}},{\"@type\":\"ListItem\",\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/investigating-error-types-in-gpt-4\\\/#listItem\",\"position\":3,\"name\":\"Investigating Error Types in GPT-4\",\"previousItem\":{\"@type\":\"ListItem\",\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/category\\\/best-publications\\\/#listItem\",\"name\":\"Best Publications\"}}]},{\"@type\":\"Organization\",\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/#organization\",\"name\":\"L3S Research Center\",\"description\":\"Trustworthy AI & Digital Transformation\",\"url\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/\",\"logo\":{\"@type\":\"ImageObject\",\"url\":\"https:\\\/\\\/www.l3s.de\\\/wp-content\\\/uploads\\\/2024\\\/11\\\/LS3-Logo-2024.png\",\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/investigating-error-types-in-gpt-4\\\/#organizationLogo\",\"width\":4188,\"height\":2549},\"image\":{\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/investigating-error-types-in-gpt-4\\\/#organizationLogo\"},\"sameAs\":[\"https:\\\/\\\/www.twitter.com\\\/l3s_luh\",\"https:\\\/\\\/www.youtube.com\\\/channel\\\/UCJzIw4-5qZJhmzOUsruHlhA\",\"https:\\\/\\\/www.linkedin.com\\\/company\\\/l3s-research-center\",\"https:\\\/\\\/de.wikipedia.org\\\/wiki\\\/L3S\"]},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/author\\\/borovynska\\\/#author\",\"url\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/author\\\/borovynska\\\/\",\"name\":\"borovynska\"},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/investigating-error-types-in-gpt-4\\\/#webpage\",\"url\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/investigating-error-types-in-gpt-4\\\/\",\"name\":\"Investigating Error Types in GPT-4 - L3S Research Center\",\"description\":\"The investigation of error types in GPT-4 is the best publication in the IR and Generative AI category.\",\"inLanguage\":\"de-DE\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/#website\"},\"breadcrumb\":{\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/investigating-error-types-in-gpt-4\\\/#breadcrumblist\"},\"author\":{\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/author\\\/borovynska\\\/#author\"},\"creator\":{\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/author\\\/borovynska\\\/#author\"},\"image\":{\"@type\":\"ImageObject\",\"url\":\"https:\\\/\\\/www.l3s.de\\\/wp-content\\\/uploads\\\/2024\\\/08\\\/IR-Generative-AI.png\",\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/investigating-error-types-in-gpt-4\\\/#mainImage\",\"width\":1200,\"height\":627},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/investigating-error-types-in-gpt-4\\\/#mainImage\"},\"datePublished\":\"2024-08-20T11:34:19+02:00\",\"dateModified\":\"2024-09-05T12:16:56+02:00\"},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/#website\",\"url\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/\",\"name\":\"L3S Research Center\",\"alternateName\":\"L3S\",\"description\":\"Trustworthy AI & Digital Transformation\",\"inLanguage\":\"de-DE\",\"publisher\":{\"@id\":\"https:\\\/\\\/www.l3s.de\\\/de\\\/#organization\"}}]}\n\t\t<\/script>\n\t\t<!-- All in One SEO -->\n\n","aioseo_head_json":{"title":"Investigating Error Types in GPT-4 - L3S Research Center","description":"Die Untersuchung der Fehlertypen in GPT-4 wurde als beste Ver\u00f6ffentlichung des L3S in der Kategorie \"IR und Generative KI\" ausgezeichnet.","canonical_url":"https:\/\/www.l3s.de\/de\/investigating-error-types-in-gpt-4\/","robots":"max-image-preview:large","keywords":"","webmasterTools":{"miscellaneous":""},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"BlogPosting","@id":"https:\/\/www.l3s.de\/de\/investigating-error-types-in-gpt-4\/#blogposting","name":"Investigating Error Types in GPT-4 - L3S Research Center","headline":"Investigating Error Types in GPT-4","author":{"@id":"https:\/\/www.l3s.de\/de\/author\/borovynska\/#author"},"publisher":{"@id":"https:\/\/www.l3s.de\/de\/#organization"},"image":{"@type":"ImageObject","url":"https:\/\/www.l3s.de\/wp-content\/uploads\/2024\/08\/IR-Generative-AI.png","width":1200,"height":627},"datePublished":"2024-08-20T11:34:19+02:00","dateModified":"2024-09-05T12:16:56+02:00","inLanguage":"de-DE","mainEntityOfPage":{"@id":"https:\/\/www.l3s.de\/de\/investigating-error-types-in-gpt-4\/#webpage"},"isPartOf":{"@id":"https:\/\/www.l3s.de\/de\/investigating-error-types-in-gpt-4\/#webpage"},"articleSection":"Best Publications"},{"@type":"BreadcrumbList","@id":"https:\/\/www.l3s.de\/de\/investigating-error-types-in-gpt-4\/#breadcrumblist","itemListElement":[{"@type":"ListItem","@id":"https:\/\/www.l3s.de\/de#listItem","position":1,"name":"Home","item":"https:\/\/www.l3s.de\/de","nextItem":{"@type":"ListItem","@id":"https:\/\/www.l3s.de\/de\/category\/best-publications\/#listItem","name":"Best Publications"}},{"@type":"ListItem","@id":"https:\/\/www.l3s.de\/de\/category\/best-publications\/#listItem","position":2,"name":"Best Publications","item":"https:\/\/www.l3s.de\/de\/category\/best-publications\/","nextItem":{"@type":"ListItem","@id":"https:\/\/www.l3s.de\/de\/investigating-error-types-in-gpt-4\/#listItem","name":"Investigating Error Types in GPT-4"},"previousItem":{"@type":"ListItem","@id":"https:\/\/www.l3s.de\/de#listItem","name":"Home"}},{"@type":"ListItem","@id":"https:\/\/www.l3s.de\/de\/investigating-error-types-in-gpt-4\/#listItem","position":3,"name":"Investigating Error Types in GPT-4","previousItem":{"@type":"ListItem","@id":"https:\/\/www.l3s.de\/de\/category\/best-publications\/#listItem","name":"Best Publications"}}]},{"@type":"Organization","@id":"https:\/\/www.l3s.de\/de\/#organization","name":"L3S Research Center","description":"Trustworthy AI & Digital Transformation","url":"https:\/\/www.l3s.de\/de\/","logo":{"@type":"ImageObject","url":"https:\/\/www.l3s.de\/wp-content\/uploads\/2024\/11\/LS3-Logo-2024.png","@id":"https:\/\/www.l3s.de\/de\/investigating-error-types-in-gpt-4\/#organizationLogo","width":4188,"height":2549},"image":{"@id":"https:\/\/www.l3s.de\/de\/investigating-error-types-in-gpt-4\/#organizationLogo"},"sameAs":["https:\/\/www.twitter.com\/l3s_luh","https:\/\/www.youtube.com\/channel\/UCJzIw4-5qZJhmzOUsruHlhA","https:\/\/www.linkedin.com\/company\/l3s-research-center","https:\/\/de.wikipedia.org\/wiki\/L3S"]},{"@type":"Person","@id":"https:\/\/www.l3s.de\/de\/author\/borovynska\/#author","url":"https:\/\/www.l3s.de\/de\/author\/borovynska\/","name":"borovynska"},{"@type":"WebPage","@id":"https:\/\/www.l3s.de\/de\/investigating-error-types-in-gpt-4\/#webpage","url":"https:\/\/www.l3s.de\/de\/investigating-error-types-in-gpt-4\/","name":"Investigating Error Types in GPT-4 - L3S Research Center","description":"The investigation of error types in GPT-4 is the best publication in the IR and Generative AI category.","inLanguage":"de-DE","isPartOf":{"@id":"https:\/\/www.l3s.de\/de\/#website"},"breadcrumb":{"@id":"https:\/\/www.l3s.de\/de\/investigating-error-types-in-gpt-4\/#breadcrumblist"},"author":{"@id":"https:\/\/www.l3s.de\/de\/author\/borovynska\/#author"},"creator":{"@id":"https:\/\/www.l3s.de\/de\/author\/borovynska\/#author"},"image":{"@type":"ImageObject","url":"https:\/\/www.l3s.de\/wp-content\/uploads\/2024\/08\/IR-Generative-AI.png","@id":"https:\/\/www.l3s.de\/de\/investigating-error-types-in-gpt-4\/#mainImage","width":1200,"height":627},"primaryImageOfPage":{"@id":"https:\/\/www.l3s.de\/de\/investigating-error-types-in-gpt-4\/#mainImage"},"datePublished":"2024-08-20T11:34:19+02:00","dateModified":"2024-09-05T12:16:56+02:00"},{"@type":"WebSite","@id":"https:\/\/www.l3s.de\/de\/#website","url":"https:\/\/www.l3s.de\/de\/","name":"L3S Research Center","alternateName":"L3S","description":"Trustworthy AI & Digital Transformation","inLanguage":"de-DE","publisher":{"@id":"https:\/\/www.l3s.de\/de\/#organization"}}]},"og:locale":"de_DE","og:site_name":"L3S Research Center - Trustworthy AI &amp; Digital Transformation","og:type":"article","og:title":"Investigating Error Types in GPT-4 - L3S Research Center","og:description":"The investigation of error types in GPT-4 is the best publication in the IR and Generative AI category.","og:url":"https:\/\/www.l3s.de\/de\/investigating-error-types-in-gpt-4\/","og:image":"https:\/\/www.l3s.de\/wp-content\/uploads\/2024\/11\/LS3-Logo-2024-WHITE.png","og:image:secure_url":"https:\/\/www.l3s.de\/wp-content\/uploads\/2024\/11\/LS3-Logo-2024-WHITE.png","article:published_time":"2024-08-20T09:34:19+00:00","article:modified_time":"2024-09-05T10:16:56+00:00","twitter:card":"summary_large_image","twitter:site":"@l3s_luh","twitter:title":"Investigating Error Types in GPT-4 - L3S Research Center","twitter:description":"The investigation of error types in GPT-4 is the best publication in the IR and Generative AI category.","twitter:creator":"@l3s_luh","twitter:image":"https:\/\/www.l3s.de\/wp-content\/uploads\/2024\/11\/LS3-Logo-2024-WHITE.png"},"aioseo_meta_data":{"post_id":"34823","title":null,"description":null,"keywords":null,"keyphrases":{"focus":{"keyphrase":"Error Types","score":70,"analysis":{"keyphraseInTitle":{"score":9,"maxScore":9,"error":0},"keyphraseInDescription":{"score":9,"maxScore":9,"error":0},"keyphraseLength":{"score":9,"maxScore":9,"error":0,"length":2},"keyphraseInURL":{"score":5,"maxScore":5,"error":0},"keyphraseInIntroduction":{"score":3,"maxScore":9,"error":1},"keyphraseInSubHeadings":[],"keyphraseInImageAlt":[],"keywordDensity":{"score":0,"type":"low","maxScore":9,"error":1}}},"additional":[]},"primary_term":null,"canonical_url":null,"og_title":null,"og_description":null,"og_object_type":"default","og_image_type":"default","og_image_url":null,"og_image_width":null,"og_image_height":null,"og_image_custom_url":null,"og_image_custom_fields":null,"og_video":"","og_custom_url":null,"og_article_section":null,"og_article_tags":null,"twitter_use_og":false,"twitter_card":"default","twitter_image_type":"default","twitter_image_url":null,"twitter_image_custom_url":null,"twitter_image_custom_fields":null,"twitter_title":null,"twitter_description":null,"schema":{"blockGraphs":[],"customGraphs":[],"default":{"data":{"Article":[],"Course":[],"Dataset":[],"FAQPage":[],"Movie":[],"Person":[],"Product":[],"ProductReview":[],"Car":[],"Recipe":[],"Service":[],"SoftwareApplication":[],"WebPage":[]},"graphName":"BlogPosting","isEnabled":true},"graphs":[]},"schema_type":"default","schema_type_options":null,"pillar_content":false,"robots_default":true,"robots_noindex":false,"robots_noarchive":false,"robots_nosnippet":false,"robots_nofollow":false,"robots_noimageindex":false,"robots_noodp":false,"robots_notranslate":false,"robots_max_snippet":"-1","robots_max_videopreview":"-1","robots_max_imagepreview":"large","priority":null,"frequency":"default","local_seo":null,"breadcrumb_settings":null,"limit_modified_date":false,"ai":null,"created":"2024-08-20 09:34:21","updated":"2025-06-16 11:11:20","seo_analyzer_scan_date":null},"aioseo_breadcrumb":"<div class=\"aioseo-breadcrumbs\"><span class=\"aioseo-breadcrumb\">\n\t\t\t<a href=\"https:\/\/www.l3s.de\/de\" title=\"Home\">Home<\/a>\n\t\t<\/span><span class=\"aioseo-breadcrumb-separator\">&raquo;<\/span><span class=\"aioseo-breadcrumb\">\n\t\t\t<a href=\"https:\/\/www.l3s.de\/de\/category\/best-publications\/\" title=\"Best Publications\">Best Publications<\/a>\n\t\t<\/span><span class=\"aioseo-breadcrumb-separator\">&raquo;<\/span><span class=\"aioseo-breadcrumb\">\n\t\t\tInvestigating Error Types in GPT-4\n\t\t<\/span><\/div>","aioseo_breadcrumb_json":[{"label":"Home","link":"https:\/\/www.l3s.de\/de"},{"label":"Best Publications","link":"https:\/\/www.l3s.de\/de\/category\/best-publications\/"},{"label":"Investigating Error Types in GPT-4","link":"https:\/\/www.l3s.de\/de\/investigating-error-types-in-gpt-4\/"}],"publishpress_future_workflow_manual_trigger":{"enabledWorkflows":[]},"_links":{"self":[{"href":"https:\/\/www.l3s.de\/de\/wp-json\/wp\/v2\/posts\/34823","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.l3s.de\/de\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.l3s.de\/de\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.l3s.de\/de\/wp-json\/wp\/v2\/users\/11"}],"replies":[{"embeddable":true,"href":"https:\/\/www.l3s.de\/de\/wp-json\/wp\/v2\/comments?post=34823"}],"version-history":[{"count":3,"href":"https:\/\/www.l3s.de\/de\/wp-json\/wp\/v2\/posts\/34823\/revisions"}],"predecessor-version":[{"id":34948,"href":"https:\/\/www.l3s.de\/de\/wp-json\/wp\/v2\/posts\/34823\/revisions\/34948"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.l3s.de\/de\/wp-json\/wp\/v2\/media\/34824"}],"wp:attachment":[{"href":"https:\/\/www.l3s.de\/de\/wp-json\/wp\/v2\/media?parent=34823"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.l3s.de\/de\/wp-json\/wp\/v2\/categories?post=34823"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.l3s.de\/de\/wp-json\/wp\/v2\/tags?post=34823"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}