{"id":15358,"date":"2024-04-09T13:09:14","date_gmt":"2024-04-09T16:09:14","guid":{"rendered":"http:\/\/fund.ar\/publicacion\/algorithmic-biases-and-social-representation-in-generative-language-models-llm\/"},"modified":"2024-04-23T12:55:24","modified_gmt":"2024-04-23T15:55:24","slug":"algorithmic-biases-and-social-representation-in-generative-language-models-llm","status":"publish","type":"publicacion","link":"https:\/\/fund.ar\/en\/publicacion\/algorithmic-biases-and-social-representation-in-generative-language-models-llm\/","title":{"rendered":"Algorithmic Biases and Social Representation in Generative Language Models (LLM)"},"content":{"rendered":"\t\t<div data-elementor-type=\"wp-post\" data-elementor-id=\"15358\" class=\"elementor elementor-15358 elementor-14335\" data-elementor-post-type=\"publicacion\">\n\t\t\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-496377c elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"496377c\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-d7f3dfa\" data-id=\"d7f3dfa\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-df7b5a1 elementor-widget elementor-widget-text-editor\" data-id=\"df7b5a1\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t<p><strong>With the emergence of GPT, generative language models (GLMs) burst into our daily lives. We use them to learn, to serve clients and even to design public policy. All the responses generated by these models (even the lack of response) reflect an opinion. Which segments of the Argentine population tend to be more present in the responses of these models? Answering this question is crucial if we want to expand the use of these tools successfully. From the lack of equal representation to the possibility of amplifying stereotypes, we analyse the challenges of these algorithmic biases.<\/strong><\/p>\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-3f06e6c elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"3f06e6c\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-a868c24\" data-id=\"a868c24\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-051afac elementor-widget elementor-widget-text-editor\" data-id=\"051afac\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t<p>Illustration: <a href=\"https:\/\/www.instagram.com\/nanni.dg\" target=\"_blank\" rel=\"noopener\">Micaela Nanni<\/a><\/p>\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-2b39a90 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"2b39a90\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-8a001db\" data-id=\"8a001db\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-e502e82 elementor-widget-divider--view-line elementor-widget elementor-widget-divider\" data-id=\"e502e82\" data-element_type=\"widget\" data-widget_type=\"divider.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<div class=\"elementor-divider\">\n\t\t\t<span class=\"elementor-divider-separator\">\n\t\t\t\t\t\t<\/span>\n\t\t<\/div>\n\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-ddee74e elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"ddee74e\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-baedd50\" data-id=\"baedd50\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-1f488c7 elementor-widget elementor-widget-heading\" data-id=\"1f488c7\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h2 class=\"elementor-heading-title elementor-size-default\">What are generative large language models (LLM) and how do they work?<\/h2>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-b2c509b elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"b2c509b\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-d03acb8\" data-id=\"d03acb8\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-22afe09 elementor-widget elementor-widget-spacer\" data-id=\"22afe09\" data-element_type=\"widget\" data-widget_type=\"spacer.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-2369064 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"2369064\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-6e9f0c9\" data-id=\"6e9f0c9\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-f5e1435 elementor-widget elementor-widget-heading\" data-id=\"f5e1435\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h4 class=\"elementor-heading-title elementor-size-default\"><b>What are the LLM models?<\/b><\/h4>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-e3d2206 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"e3d2206\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-58f0f2e\" data-id=\"58f0f2e\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-1ff05f8 elementor-widget elementor-widget-text-editor\" data-id=\"1ff05f8\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t<p>They are natural language processing models, designed to interpret and answer questions in increasingly sophisticated ways. They emerged in academia, but it was the arrival of ChatGPT in November 2022 that spurred their expansion into diverse areas of society.<\/p>\n<p>Since then, <a href=\"https:\/\/fund.ar\/publicacion\/ia-generativa-una-cuestion-politica\/\" target=\"_blank\" rel=\"noopener\">these models are finding applications in a variety of domains<\/a>, from customer care or personalisation of recommendations to evidence-based public policy design. However, not all of their uses are virtuous. There have also been negative cases, for example in the use in governmental contexts for distorting information and generating misleading discourse to promote political or ideological agendas.<\/p>\n<p>As these tools become more and more massive, responsibility for their use and understanding their ethical implications has also become an issue in itself. A fundamental question arises: to what extent do these algorithms reflect and may even amplify pre-existing biases in our society?<\/p>\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-15428d7 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"15428d7\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-db4be85\" data-id=\"db4be85\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-9d0c316 elementor-widget elementor-widget-spacer\" data-id=\"9d0c316\" data-element_type=\"widget\" data-widget_type=\"spacer.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-d501f7b elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"d501f7b\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-62b5d35\" data-id=\"62b5d35\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-43a2311 elementor-widget elementor-widget-heading\" data-id=\"43a2311\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h4 class=\"elementor-heading-title elementor-size-default\"><b>How do the LLM models work?<\/b><\/h4>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-92f815b elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"92f815b\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-be0911c\" data-id=\"be0911c\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-9f3b07b elementor-widget-tablet_extra__width-initial elementor-widget elementor-widget-text-editor\" data-id=\"9f3b07b\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t<p>They work through <i>prompts<\/i>: prompts are given to generate specific responses or content. In other words, a <i>promp<\/i>t is a question asked of the model. The model, based on the vast amount of information it has been trained with, produces a coherent and, in many cases, surprisingly accurate answer.<\/p><p>This training takes place in two stages:<\/p><ul><li><b>Training<\/b>, where the model develops essential skills, such as understanding and generating language. The process is resource-intensive and only technology giants such as Google, Meta and OpenAI have successfully implemented it. The rest benefit from models pre-trained by these companies.<\/li><li><b>Fine-tuning<\/b>, where the behaviour of each model is defined: their conversational tone, limits on their responses and friendliness, among others. This phase is less resource-intensive but demands greater human involvement to ensure that responses are aligned with specific standards.<\/li><li style=\"list-style-type: none;\"><ul><li style=\"list-style-type: none;\">\u00a0<\/li><\/ul><\/li><\/ul><p>In this fine-tuning, models may inherit biases due to human intervention that seeks to intentionally direct their behaviour. It is these biases that will subsequently cause them to answer questions based on the beliefs with which they were trained.<\/p>\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-c797568 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"c797568\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-ae8fd6c\" data-id=\"ae8fd6c\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-51f4c87 elementor-widget elementor-widget-spacer\" data-id=\"51f4c87\" data-element_type=\"widget\" data-widget_type=\"spacer.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-722ac5a elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"722ac5a\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-4317155\" data-id=\"4317155\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-343885a elementor-widget elementor-widget-heading\" data-id=\"343885a\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h4 class=\"elementor-heading-title elementor-size-default\"><b>How do LLM models \u2018get it wrong\u2019?<\/b><\/h4>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-d6a9f39 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"d6a9f39\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-2b709b7\" data-id=\"2b709b7\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-ee2072b elementor-widget elementor-widget-text-editor\" data-id=\"ee2072b\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t<p>They generate text by predicting the next most likely word, based on the data used during their training. They tend to express themselves with a high degree of confidence. The combination of these two characteristics can result in answers that are inaccurate or even completely fictitious, but sound authentic. Texts produced by LLMs can have the following drawbacks:\u00a0<\/p><ul><li aria-level=\"1\"><b>Biases <\/b>(gender, race, sexual orientation or other attributes): an incorrect or unfair representation of a population or phenomenon, given by partial or incorrect data collection or by already existing biases (present in the data collected).<\/li><li aria-level=\"1\"><b>Misleading or false information<\/b>: this can happen if the training data contains inaccurate information. This is especially problematic when used for consultations that require accurate information, such as medical advice.<\/li><li aria-level=\"1\"><b>Hallucinations<\/b>: models may generate fictitious or fabricated responses, which can be misleading or potentially harmful if users take such information as true.<\/li><li aria-level=\"1\"><b>Reinforcement of pre-existing beliefs<\/b>: which could lead to polarisation and lack of diversity of opinion.<\/li><\/ul>\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-c6d1f7c elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"c6d1f7c\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-05af662\" data-id=\"05af662\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-7b222a7 elementor-widget elementor-widget-spacer\" data-id=\"7b222a7\" data-element_type=\"widget\" data-widget_type=\"spacer.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-69ed98b elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"69ed98b\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-e7873aa\" data-id=\"e7873aa\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-667e3b2 elementor-widget elementor-widget-heading\" data-id=\"667e3b2\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h2 class=\"elementor-heading-title elementor-size-default\">Whose views reflect GPT and its competitors?<\/h2>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-aac96ba elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"aac96ba\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-1081015\" data-id=\"1081015\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-f7037ed elementor-widget elementor-widget-spacer\" data-id=\"f7037ed\" data-element_type=\"widget\" data-widget_type=\"spacer.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-0970776 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"0970776\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-588e9ed\" data-id=\"588e9ed\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-58901d8 elementor-widget elementor-widget-heading\" data-id=\"58901d8\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h4 class=\"elementor-heading-title elementor-size-default\"><b>Research methodology<\/b><\/h4>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-3cf4046 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"3cf4046\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-887e7c7\" data-id=\"887e7c7\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-635f03d elementor-widget elementor-widget-text-editor\" data-id=\"635f03d\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t<p><span>This research is based on <a href=\"https:\/\/arxiv.org\/pdf\/2303.17548.pdf\" target=\"_blank\" rel=\"noopener\">a study conducted by Stanford University<\/a>, which compared the responses of various generative language models (LLM), including GPT, with the opinions of US society through public opinion polls.\u00a0<\/span><\/p>\n<p><span>For this study:\u00a0<\/span><\/p>\n<ol>\n<li aria-level=\"1\"><b>We selected a set of 78 questions and<\/b><span> answers from the Latinobar\u00f3metro opinion poll (2020).<\/span><\/li>\n<li aria-level=\"1\"><b>We repeated the questionnaire<\/b><span>: we asked the same survey questions to three LLM models: <a href=\"https:\/\/chat.openai.com\/auth\/login\" target=\"_blank\" rel=\"noopener\">GPT Turbo 3.5<\/a> (OpenAI),\u00a0<a href=\"https:\/\/cohere.com\/models\/command\" target=\"_blank\" rel=\"noopener\">Command-nightly<\/a> (Cohere), Bard (older version of\u00a0<a href=\"https:\/\/gemini.google.com\/?hl=es\" target=\"_blank\" rel=\"noopener\">Gemini<\/a>) (Google).\u00a0<\/span><\/li>\n<li aria-level=\"1\"><b>We compare the responses <\/b>of<span>\u00a0these 3 LLM models with those of the Argentinean population<\/span>.<\/li>\n<li aria-level=\"1\"><b>We analysed commonalities and differences <\/b><span>to identify the characteristics that each model shares with different segments of the local population.\u00a0<\/span><\/li>\n<\/ol>\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-6f706f3 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"6f706f3\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-8639f2a\" data-id=\"8639f2a\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-5285925 elementor-widget elementor-widget-spacer\" data-id=\"5285925\" data-element_type=\"widget\" data-widget_type=\"spacer.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-058fccd elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"058fccd\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-daf14a2\" data-id=\"daf14a2\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-240ad97 elementor-widget elementor-widget-heading\" data-id=\"240ad97\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h4 class=\"elementor-heading-title elementor-size-default\"><b>Which segments of the population tend to be more present in the responses of LLM models?<\/b><\/h4>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-4e2c67e elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"4e2c67e\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-ba64deb\" data-id=\"ba64deb\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-d0aab93 elementor-widget elementor-widget-text-editor\" data-id=\"d0aab93\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t<p>When analysing the results, we observed that people who showed similarity in their responses to each Model (LLM) had the following characteristics:<\/p>\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-93bf8bd elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"93bf8bd\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-3c5ea40\" data-id=\"3c5ea40\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-de68531 elementor-widget elementor-widget-text-editor\" data-id=\"de68531\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t<table>\n<thead>\n<tr>\n<th rowspan=\"2\"><\/th>\n<th colspan=\"3\">\n<h6><b>LLM MODELS ANALYSED<\/b><\/h6>\n<\/th>\n<\/tr>\n<tr>\n<th>\n<h6><b>GPT Turbo 3.5<\/b><\/h6>\n<\/th>\n<th>\n<h6><b>Cohere<\/b><\/h6>\n<\/th>\n<th>\n<h6><b>Bard<\/b><\/h6>\n<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>\n<h6><b>PROFILE<\/b><\/h6>\n<p><span>Characteristics<\/span><b>\u00a0<\/b><span>of the people with whom they showed the greatest similarity in their responses<\/span>.<\/td>\n<td><span>Male<\/span><\/p>\n<p><span>Interest in politics<\/span><\/p>\n<p><span>Adult<\/span><\/p>\n<p><span>High educational level<\/span><\/p>\n<p><span>Right-leaning ideology<\/span><\/td>\n<td><span>Male<\/span><\/p>\n<p><span>Interest in politics<\/span><\/td>\n<td><span>Male<\/span><\/p>\n<p><span>Interest in politics<\/span><\/p>\n<p><span>Adult<\/span><\/p>\n<p><span>High educational level<\/span><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-bb5098a elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"bb5098a\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-200c7db\" data-id=\"200c7db\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-62158d9 elementor-widget elementor-widget-heading\" data-id=\"62158d9\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h6 class=\"elementor-heading-title elementor-size-default\"><b>Profile: Male<\/b><\/h6>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-5e7d06c elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"5e7d06c\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-abf21a6\" data-id=\"abf21a6\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-43295ea elementor-widget elementor-widget-image\" data-id=\"43295ea\" data-element_type=\"widget\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<img fetchpriority=\"high\" decoding=\"async\" width=\"800\" height=\"783\" src=\"https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_GraficosWeb_Sesgos_Grafico2-1024x1002.png\" class=\"attachment-large size-large wp-image-14692\" alt=\"\" srcset=\"https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_GraficosWeb_Sesgos_Grafico2-1024x1002.png 1024w, https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_GraficosWeb_Sesgos_Grafico2-300x293.png 300w, https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_GraficosWeb_Sesgos_Grafico2-768x751.png 768w, https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_GraficosWeb_Sesgos_Grafico2-1536x1502.png 1536w, https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_GraficosWeb_Sesgos_Grafico2.png 1821w\" sizes=\"(max-width: 800px) 100vw, 800px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-b3d5ea2 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"b3d5ea2\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-ce2abd2\" data-id=\"ce2abd2\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-50337cb elementor-widget elementor-widget-heading\" data-id=\"50337cb\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h6 class=\"elementor-heading-title elementor-size-default\"><b>Profile: High educational level<\/b><\/h6>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-0ea4f78 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"0ea4f78\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-4383f38\" data-id=\"4383f38\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-fd67357 elementor-widget elementor-widget-image\" data-id=\"fd67357\" data-element_type=\"widget\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<img decoding=\"async\" width=\"800\" height=\"750\" src=\"https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_GraficosWeb_Sesgos_Grafico8-1024x960.png\" class=\"attachment-large size-large wp-image-14694\" alt=\"\" srcset=\"https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_GraficosWeb_Sesgos_Grafico8-1024x960.png 1024w, https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_GraficosWeb_Sesgos_Grafico8-300x281.png 300w, https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_GraficosWeb_Sesgos_Grafico8-768x720.png 768w, https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_GraficosWeb_Sesgos_Grafico8-1536x1441.png 1536w, https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_GraficosWeb_Sesgos_Grafico8.png 1821w\" sizes=\"(max-width: 800px) 100vw, 800px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-73baa03 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"73baa03\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-4a289eb\" data-id=\"4a289eb\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-a0055ee elementor-widget elementor-widget-heading\" data-id=\"a0055ee\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h6 class=\"elementor-heading-title elementor-size-default\"><b>Profile: Interest in politics<\/b><\/h6>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-66047d6 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"66047d6\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-81b6151\" data-id=\"81b6151\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-ba913be elementor-widget elementor-widget-image\" data-id=\"ba913be\" data-element_type=\"widget\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<img decoding=\"async\" width=\"800\" height=\"750\" src=\"https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_GraficosWeb_Sesgos_Grafico7-1024x960.png\" class=\"attachment-large size-large wp-image-14696\" alt=\"\" srcset=\"https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_GraficosWeb_Sesgos_Grafico7-1024x960.png 1024w, https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_GraficosWeb_Sesgos_Grafico7-300x281.png 300w, https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_GraficosWeb_Sesgos_Grafico7-768x720.png 768w, https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_GraficosWeb_Sesgos_Grafico7-1536x1441.png 1536w, https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_GraficosWeb_Sesgos_Grafico7.png 1821w\" sizes=\"(max-width: 800px) 100vw, 800px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-a202c6b elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"a202c6b\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-f0ba932\" data-id=\"f0ba932\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-8d09430 elementor-widget elementor-widget-heading\" data-id=\"8d09430\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h6 class=\"elementor-heading-title elementor-size-default\"><b>Profile: Right-leaning ideology<\/b><\/h6>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-2cd3601 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"2cd3601\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-19a951f\" data-id=\"19a951f\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-34417ea elementor-widget elementor-widget-image\" data-id=\"34417ea\" data-element_type=\"widget\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<img loading=\"lazy\" decoding=\"async\" width=\"800\" height=\"805\" src=\"https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_GraficosWeb_Sesgos_Grafico3-1018x1024.png\" class=\"attachment-large size-large wp-image-14698\" alt=\"\" srcset=\"https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_GraficosWeb_Sesgos_Grafico3-1018x1024.png 1018w, https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_GraficosWeb_Sesgos_Grafico3-298x300.png 298w, https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_GraficosWeb_Sesgos_Grafico3-150x150.png 150w, https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_GraficosWeb_Sesgos_Grafico3-768x772.png 768w, https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_GraficosWeb_Sesgos_Grafico3-1528x1536.png 1528w, https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_GraficosWeb_Sesgos_Grafico3.png 1821w\" sizes=\"(max-width: 800px) 100vw, 800px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-b72c921 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"b72c921\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-d04298d\" data-id=\"d04298d\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-bc8f1ad elementor-widget elementor-widget-spacer\" data-id=\"bc8f1ad\" data-element_type=\"widget\" data-widget_type=\"spacer.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-7ab29c6 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"7ab29c6\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-61968d7\" data-id=\"61968d7\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-ddea5ce elementor-widget elementor-widget-heading\" data-id=\"ddea5ce\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h5 class=\"elementor-heading-title elementor-size-default\">LLM models characterisation<\/h5>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-59e4812 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"59e4812\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-e5a24f4\" data-id=\"e5a24f4\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-27d2ae4 elementor-widget elementor-widget-text-editor\" data-id=\"27d2ae4\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t<p>Although the responses of the 3 models are not identical, they share certain characteristics that make them similar in terms of audiences. All three models show a bias towards a more male and politicised sector. In addition, both Bard and GPT Turbo 3.5 also resemble people with higher levels of education and older age. It is worth noting that GPT Turbo 3.5 is the only one that shows a significant correlation with ideology, showing more affinity with individuals with right-wing ideological orientations.<\/p>\n<ul>\n<li><strong>The affinity of these models with people interested in politics<\/strong> makes sense, since, during their training, they were exposed to a variety of policy-related data sources, such as social media discussions, Wikipedia definitions and essays.<\/li>\n<li><strong>The overlap with highly educated individuals<\/strong> is understandable, given that these models are very knowledgeable due to their training.<\/li>\n<li><strong>The trend towards masculinisation<\/strong> can be explained by the fact that, even if we do not know their exact identities, the software field tends to be dominated by men and that the majority of the authors of the founding LLM papers (13 out of 16) are men.<\/li>\n<\/ul>\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-febb79d elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"febb79d\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-f8c0f8c\" data-id=\"f8c0f8c\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-832a2b7 elementor-widget elementor-widget-spacer\" data-id=\"832a2b7\" data-element_type=\"widget\" data-widget_type=\"spacer.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-13cd513 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"13cd513\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-35835a6\" data-id=\"35835a6\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-14de69a elementor-widget elementor-widget-heading\" data-id=\"14de69a\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h4 class=\"elementor-heading-title elementor-size-default\"><b>Profiles change according to the topics on which they are consulted<\/b><\/h4>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-d4996ce elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"d4996ce\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-a0ab51b\" data-id=\"a0ab51b\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-ec59680 elementor-widget elementor-widget-text-editor\" data-id=\"ec59680\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">We then analyzed by topic, considering how the profiles vary according to the topics on which they are queried. Since some LLM did not answer many of these questions, we indicated &#8220;No data&#8221; when the model did not choose any of the alternatives provided.<\/span><\/p>\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-7c8ddda elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"7c8ddda\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-7854be8\" data-id=\"7854be8\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-58e93e8 elementor-widget elementor-widget-text-editor\" data-id=\"58e93e8\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t<table>\n<tbody>\n<tr>\n<td><b>Topic<\/b><\/td>\n<td><b>GPT Turbo 3.5<\/b><\/td>\n<td><b>Cohere<\/b><\/td>\n<td><b>Bard<\/b><\/td>\n<\/tr>\n<tr>\n<td><span>International Relations<\/span><\/td>\n<td><span>No data<\/span><\/td>\n<td><span>No data<\/span><\/td>\n<td><span>&#8211; Adult<\/span><\/p>\n<p><span>&#8211; Male<\/span><\/p>\n<p><span>&#8211; Right-leaning ideology<\/span><\/p>\n<p><span>&#8211; High educational level<\/span><\/p>\n<p><span>&#8211; Politicised<\/span><\/td>\n<\/tr>\n<tr>\n<td><span>Opinion\u00a0<\/span><\/td>\n<td><span>No data<\/span><\/td>\n<td><span>No data<\/span><\/td>\n<td><span>&#8211; High educational level<\/span><\/p>\n<p><span>&#8211; Do not emigrate<\/span><span><br \/>\n<\/span><span>&#8211; Politicised<\/span>.<\/td>\n<\/tr>\n<tr>\n<td><span>Economy\u00a0<\/span><\/td>\n<td><span>No data<\/span><\/td>\n<td><span>No data<\/span><\/td>\n<td><span>&#8211; Adult<\/span><\/p>\n<p><span>&#8211; Male<\/span><\/p>\n<p><span>&#8211; Right-leaning ideology<\/span><\/p>\n<p><span>&#8211; High educational level<\/span><\/p>\n<p><span>&#8211; Desire to emigrate<\/span><\/p>\n<p><span>&#8211; Politicised<\/span><\/td>\n<\/tr>\n<tr>\n<td><span>Democracy\u00a0<\/span><\/td>\n<td><span>No data<\/span><\/td>\n<td><span>No data<\/span><\/td>\n<td><span>&#8211; Left-leaning ideology<\/span><\/p>\n<p><span>&#8211; Do not emigrate<\/span><\/p>\n<p><span>&#8211; Politicised<\/span><\/td>\n<\/tr>\n<tr>\n<td><span>Ideology\u00a0<\/span><\/td>\n<td><span>No data<\/span><\/td>\n<td><span>No data<\/span><\/td>\n<td><span>&#8211; Male<\/span><\/p>\n<p><span>&#8211; Desire to emigrate<\/span><\/td>\n<\/tr>\n<tr>\n<td><span>Social rights<\/span><\/td>\n<td><span>&#8211; Politicised<\/span><\/p>\n<p><span>&#8211; Adult<\/span><\/td>\n<td><span>&#8211; Politicised<\/span><span><br \/>\n<\/span><span>&#8211; Adult<\/span>.<\/p>\n<p><span>&#8211; Do not emigrate<\/span><\/td>\n<td><span>&#8211; Politicised<\/span><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-d0a4a8f elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"d0a4a8f\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-98f4239\" data-id=\"98f4239\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-e65a152 elementor-widget elementor-widget-heading\" data-id=\"e65a152\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h5 class=\"elementor-heading-title elementor-size-default\">Characterisation of LLM models according to topics<\/h5>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-49fd5ad elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"49fd5ad\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-a29602c\" data-id=\"a29602c\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-a286f11 elementor-widget elementor-widget-text-editor\" data-id=\"a286f11\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t<p>While there are similarities on several issues, we observe that the profiles of the models vary according to the topics asked about. For example, Bard has an affinity with a right-wing audience when asked about international relations issues, but more closely resembles a left-wing audience when asked about democracy. The same is true for his perspective on the desire to emigrate from the country. When asked about economics or ideology, they seem to have more similarities with those who wish to leave the country, but when it comes to questions related to democracy or opinion, the opposite is true.<\/p>\n<p>Both GPT Turbo 3.5 and Cohere had a high number of unanswered questions: 53 out of 78 for GPT (68% unanswered) and 50 for Cohere (64%). This indicates an effort on the part of the developers to prevent these models from issuing opinions on numerous issues. In contrast, Bard refused to answer only 11 questions (only 14% unanswered), suggesting a less restrictive approach by Google.<\/p>\n<p>In terms of bias, it is notable that both Bard and Cohere responded by referring to the United States when asked about &#8220;our country&#8221;. In contrast, GPT Turbo 3.5 answered that they did not know which country was being referred to. This difference highlights how developer biases may have influenced responses in possibly unintentional ways.<\/p>\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-2a04cb4 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"2a04cb4\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-058c371\" data-id=\"058c371\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-51c2750 elementor-widget elementor-widget-spacer\" data-id=\"51c2750\" data-element_type=\"widget\" data-widget_type=\"spacer.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-99f5999 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"99f5999\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-600796e\" data-id=\"600796e\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-c9f387d elementor-widget elementor-widget-heading\" data-id=\"c9f387d\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h2 class=\"elementor-heading-title elementor-size-default\">Good practices for the use and design of LLM models<\/h2>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-6c7e6c3 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"6c7e6c3\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-3bd2830\" data-id=\"3bd2830\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-1f13b8a elementor-widget elementor-widget-text-editor\" data-id=\"1f13b8a\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">The complete elimination of biases in general-purpose models, such as LLMs, remains a challenge without a definitive solution at present. What is currently done is to orient the behaviour of models towards what their developers consider to be the &#8220;good&#8221;. But this notion of &#8220;good&#8221; often depends on the culture and context in which it is developed. It is therefore of utmost importance to identify these biases to use this technology more responsibly and effectively, recognising its limitations and considering how it may affect various audiences and applications.<\/span><\/p>\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-065fed1 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"065fed1\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-4aaf1a7\" data-id=\"4aaf1a7\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-b52a9ee elementor-widget elementor-widget-spacer\" data-id=\"b52a9ee\" data-element_type=\"widget\" data-widget_type=\"spacer.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-7b28769 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"7b28769\" data-element_type=\"section\" data-settings=\"{&quot;background_background&quot;:&quot;classic&quot;}\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-bda19bb\" data-id=\"bda19bb\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-1f13107 elementor-widget elementor-widget-text-editor\" data-id=\"1f13107\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t<p>To reproduce the analysis performed in this study, you can access the repository available <a href=\"https:\/\/github.com\/datos-Fundar\/sesgos_LLM\" target=\"_blank\" rel=\"noopener\">at the following link<\/a>. This repository contains all the data and scripts necessary to replicate the procedures presented in this work. It is important to note that the code present in the repository reflects the functionality available at the time of the creation of this document. Due to possible changes in APIs or other factors, there may be variations in the results if executed at a later time.<\/p>\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<\/div>\n\t\t","protected":false},"featured_media":14570,"template":"","meta":{"_acf_changed":false,"_links_to":"","_links_to_target":""},"tipo":[154],"serie":[206],"desafios":[129,249,247],"agendas":[262,9,234],"class_list":["post-15358","publicacion","type-publicacion","status-publish","has-post-thumbnail","hentry","tipo-working-document","serie-artificial-intelligence","desafios-bienestar-compartido","desafios-shared-well-being","desafios-smart-governance","agendas-ai-for-development","agendas-datos-para-el-desarrollo","agendas-ia-para-el-desarrollo"],"acf":[],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v26.5 - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>Biases and social representation in ChatGPT and LLM models<\/title>\n<meta name=\"description\" content=\"Which segments of the Argentinean population tend to be more present in the responses of these LLM models?\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/fund.ar\/en\/publicacion\/algorithmic-biases-and-social-representation-in-generative-language-models-llm\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Biases and social representation in ChatGPT and LLM models\" \/>\n<meta property=\"og:description\" content=\"Which segments of the Argentinean population tend to be more present in the responses of these LLM models?\" \/>\n<meta property=\"og:url\" content=\"https:\/\/fund.ar\/en\/publicacion\/algorithmic-biases-and-social-representation-in-generative-language-models-llm\/\" \/>\n<meta property=\"og:site_name\" content=\"Fundar\" \/>\n<meta property=\"article:modified_time\" content=\"2024-04-23T15:55:24+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_Portada_Sesgos.jpg\" \/>\n\t<meta property=\"og:image:width\" content=\"1920\" \/>\n\t<meta property=\"og:image:height\" content=\"1080\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/jpeg\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:site\" content=\"@fundesarg\" \/>\n<meta name=\"twitter:label1\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data1\" content=\"8 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/fund.ar\/en\/publicacion\/algorithmic-biases-and-social-representation-in-generative-language-models-llm\/\",\"url\":\"https:\/\/fund.ar\/en\/publicacion\/algorithmic-biases-and-social-representation-in-generative-language-models-llm\/\",\"name\":\"Biases and social representation in ChatGPT and LLM models\",\"isPartOf\":{\"@id\":\"https:\/\/fund.ar\/en\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\/\/fund.ar\/en\/publicacion\/algorithmic-biases-and-social-representation-in-generative-language-models-llm\/#primaryimage\"},\"image\":{\"@id\":\"https:\/\/fund.ar\/en\/publicacion\/algorithmic-biases-and-social-representation-in-generative-language-models-llm\/#primaryimage\"},\"thumbnailUrl\":\"https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_Portada_Sesgos.jpg\",\"datePublished\":\"2024-04-09T16:09:14+00:00\",\"dateModified\":\"2024-04-23T15:55:24+00:00\",\"description\":\"Which segments of the Argentinean population tend to be more present in the responses of these LLM models?\",\"breadcrumb\":{\"@id\":\"https:\/\/fund.ar\/en\/publicacion\/algorithmic-biases-and-social-representation-in-generative-language-models-llm\/#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/fund.ar\/en\/publicacion\/algorithmic-biases-and-social-representation-in-generative-language-models-llm\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/fund.ar\/en\/publicacion\/algorithmic-biases-and-social-representation-in-generative-language-models-llm\/#primaryimage\",\"url\":\"https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_Portada_Sesgos.jpg\",\"contentUrl\":\"https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_Portada_Sesgos.jpg\",\"width\":1920,\"height\":1080},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/fund.ar\/en\/publicacion\/algorithmic-biases-and-social-representation-in-generative-language-models-llm\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Proyectos\",\"item\":\"https:\/\/fund.ar\/en\/proyectos\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Algorithmic Biases and Social Representation in Generative Language Models (LLM)\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/fund.ar\/en\/#website\",\"url\":\"https:\/\/fund.ar\/en\/\",\"name\":\"Fundar\",\"description\":\"Fundar\",\"publisher\":{\"@id\":\"https:\/\/fund.ar\/en\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/fund.ar\/en\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"},{\"@type\":\"Organization\",\"@id\":\"https:\/\/fund.ar\/en\/#organization\",\"name\":\"Fundar\",\"url\":\"https:\/\/fund.ar\/en\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/fund.ar\/en\/#\/schema\/logo\/image\/\",\"url\":\"https:\/\/fund.ar\/wp-content\/uploads\/2022\/07\/logo_fundar-light.svg\",\"contentUrl\":\"https:\/\/fund.ar\/wp-content\/uploads\/2022\/07\/logo_fundar-light.svg\",\"width\":175,\"height\":90,\"caption\":\"Fundar\"},\"image\":{\"@id\":\"https:\/\/fund.ar\/en\/#\/schema\/logo\/image\/\"},\"sameAs\":[\"https:\/\/x.com\/fundesarg\"]}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"Biases and social representation in ChatGPT and LLM models","description":"Which segments of the Argentinean population tend to be more present in the responses of these LLM models?","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/fund.ar\/en\/publicacion\/algorithmic-biases-and-social-representation-in-generative-language-models-llm\/","og_locale":"en_US","og_type":"article","og_title":"Biases and social representation in ChatGPT and LLM models","og_description":"Which segments of the Argentinean population tend to be more present in the responses of these LLM models?","og_url":"https:\/\/fund.ar\/en\/publicacion\/algorithmic-biases-and-social-representation-in-generative-language-models-llm\/","og_site_name":"Fundar","article_modified_time":"2024-04-23T15:55:24+00:00","og_image":[{"width":1920,"height":1080,"url":"https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_Portada_Sesgos.jpg","type":"image\/jpeg"}],"twitter_card":"summary_large_image","twitter_site":"@fundesarg","twitter_misc":{"Est. reading time":"8 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/fund.ar\/en\/publicacion\/algorithmic-biases-and-social-representation-in-generative-language-models-llm\/","url":"https:\/\/fund.ar\/en\/publicacion\/algorithmic-biases-and-social-representation-in-generative-language-models-llm\/","name":"Biases and social representation in ChatGPT and LLM models","isPartOf":{"@id":"https:\/\/fund.ar\/en\/#website"},"primaryImageOfPage":{"@id":"https:\/\/fund.ar\/en\/publicacion\/algorithmic-biases-and-social-representation-in-generative-language-models-llm\/#primaryimage"},"image":{"@id":"https:\/\/fund.ar\/en\/publicacion\/algorithmic-biases-and-social-representation-in-generative-language-models-llm\/#primaryimage"},"thumbnailUrl":"https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_Portada_Sesgos.jpg","datePublished":"2024-04-09T16:09:14+00:00","dateModified":"2024-04-23T15:55:24+00:00","description":"Which segments of the Argentinean population tend to be more present in the responses of these LLM models?","breadcrumb":{"@id":"https:\/\/fund.ar\/en\/publicacion\/algorithmic-biases-and-social-representation-in-generative-language-models-llm\/#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/fund.ar\/en\/publicacion\/algorithmic-biases-and-social-representation-in-generative-language-models-llm\/"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/fund.ar\/en\/publicacion\/algorithmic-biases-and-social-representation-in-generative-language-models-llm\/#primaryimage","url":"https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_Portada_Sesgos.jpg","contentUrl":"https:\/\/fund.ar\/wp-content\/uploads\/2024\/03\/FU_Portada_Sesgos.jpg","width":1920,"height":1080},{"@type":"BreadcrumbList","@id":"https:\/\/fund.ar\/en\/publicacion\/algorithmic-biases-and-social-representation-in-generative-language-models-llm\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Proyectos","item":"https:\/\/fund.ar\/en\/proyectos\/"},{"@type":"ListItem","position":2,"name":"Algorithmic Biases and Social Representation in Generative Language Models (LLM)"}]},{"@type":"WebSite","@id":"https:\/\/fund.ar\/en\/#website","url":"https:\/\/fund.ar\/en\/","name":"Fundar","description":"Fundar","publisher":{"@id":"https:\/\/fund.ar\/en\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/fund.ar\/en\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":"Organization","@id":"https:\/\/fund.ar\/en\/#organization","name":"Fundar","url":"https:\/\/fund.ar\/en\/","logo":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/fund.ar\/en\/#\/schema\/logo\/image\/","url":"https:\/\/fund.ar\/wp-content\/uploads\/2022\/07\/logo_fundar-light.svg","contentUrl":"https:\/\/fund.ar\/wp-content\/uploads\/2022\/07\/logo_fundar-light.svg","width":175,"height":90,"caption":"Fundar"},"image":{"@id":"https:\/\/fund.ar\/en\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/x.com\/fundesarg"]}]}},"_links":{"self":[{"href":"https:\/\/fund.ar\/en\/wp-json\/wp\/v2\/publicacion\/15358","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/fund.ar\/en\/wp-json\/wp\/v2\/publicacion"}],"about":[{"href":"https:\/\/fund.ar\/en\/wp-json\/wp\/v2\/types\/publicacion"}],"version-history":[{"count":25,"href":"https:\/\/fund.ar\/en\/wp-json\/wp\/v2\/publicacion\/15358\/revisions"}],"predecessor-version":[{"id":15392,"href":"https:\/\/fund.ar\/en\/wp-json\/wp\/v2\/publicacion\/15358\/revisions\/15392"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/fund.ar\/en\/wp-json\/wp\/v2\/media\/14570"}],"wp:attachment":[{"href":"https:\/\/fund.ar\/en\/wp-json\/wp\/v2\/media?parent=15358"}],"wp:term":[{"taxonomy":"tipo","embeddable":true,"href":"https:\/\/fund.ar\/en\/wp-json\/wp\/v2\/tipo?post=15358"},{"taxonomy":"serie","embeddable":true,"href":"https:\/\/fund.ar\/en\/wp-json\/wp\/v2\/serie?post=15358"},{"taxonomy":"desafios","embeddable":true,"href":"https:\/\/fund.ar\/en\/wp-json\/wp\/v2\/desafios?post=15358"},{"taxonomy":"agendas","embeddable":true,"href":"https:\/\/fund.ar\/en\/wp-json\/wp\/v2\/agendas?post=15358"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}