{"id":6320,"date":"2023-02-06T09:00:28","date_gmt":"2023-02-06T08:00:28","guid":{"rendered":"https:\/\/www.orizoconsult.com\/?p=6320"},"modified":"2023-04-27T17:28:50","modified_gmt":"2023-04-27T15:28:50","slug":"cutting-edge-delay-analysis-series-chatgpt-technologies-disruption-forensic-industry-part-1-embeddings","status":"publish","type":"post","link":"https:\/\/www.orizoconsult.com\/insights\/chatgpt-technologies-disruption-forensic-industry-part-1-embeddings\/","title":{"rendered":"Cutting Edge Delay Analysis Series &#8211; ChatGPT technologies: a disruption to the forensic industry? \u2013 Part 1: Embeddings"},"content":{"rendered":"\t\t<div data-elementor-type=\"wp-post\" data-elementor-id=\"6320\" class=\"elementor elementor-6320\" data-elementor-post-type=\"post\">\n\t\t\t\t\t\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-49b31a8 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"49b31a8\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-no\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-fed9b68\" data-id=\"fed9b68\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-1dc570e elementor-widget elementor-widget-image\" data-id=\"1dc570e\" 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<figure class=\"wp-caption\">\n\t\t\t\t\t\t\t\t\t\t<img decoding=\"async\" src=\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/elementor\/thumbs\/pexels-pixabay-267669-scaled-q28plz68wnraeifhcaax1sjngbh3jzmix3194zpyfc.jpg\" title=\"pexels-pixabay-267669\" alt=\"pexels-pixabay-267669\" loading=\"lazy\" \/>\t\t\t\t\t\t\t\t\t\t\t<figcaption class=\"widget-image-caption wp-caption-text\">ChatGPT and the future of the forensic industry \u2013 Part 1: Embeddings<\/figcaption>\n\t\t\t\t\t\t\t\t\t\t<\/figure>\n\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-2c02592 elementor-widget elementor-widget-spacer\" data-id=\"2c02592\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<section class=\"elementor-section elementor-inner-section elementor-element elementor-element-c22aca4 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"c22aca4\" 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-50 elementor-inner-column elementor-element elementor-element-ff19c18\" data-id=\"ff19c18\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-1020b67 elementor-align-center elementor-widget elementor-widget-button\" data-id=\"1020b67\" data-element_type=\"widget\" data-widget_type=\"button.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"elementor-button-wrapper\">\n\t\t\t<a class=\"elementor-button elementor-button-link elementor-size-sm\" href=\"https:\/\/www.linkedin.com\/newsletters\/7021095885036003328\/\" target=\"_blank\">\n\t\t\t\t\t\t<span class=\"elementor-button-content-wrapper\">\n\t\t\t\t\t\t<span class=\"elementor-button-icon elementor-align-icon-left\">\n\t\t\t\t<i aria-hidden=\"true\" class=\"fab fa-linkedin\"><\/i>\t\t\t<\/span>\n\t\t\t\t\t\t<span class=\"elementor-button-text\">Visit Newsletter<\/span>\n\t\t<\/span>\n\t\t\t\t\t<\/a>\n\t\t<\/div>\n\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<div class=\"elementor-column elementor-col-50 elementor-inner-column elementor-element elementor-element-bb45516\" data-id=\"bb45516\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-faf7bdf elementor-widget elementor-widget-text-editor\" data-id=\"faf7bdf\" 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\tThis article is part of the Cutting Edge Delay Analysis series, which is also available as a LinkedIn newsletter.\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\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<div class=\"elementor-element elementor-element-550eb23 elementor-widget elementor-widget-spacer\" data-id=\"550eb23\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-73a1198 elementor-widget elementor-widget-text-editor\" data-id=\"73a1198\" 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<p><em>On 30 November 2022, OpenAI released its latest chat bot technology: ChatGPT. With its natural and human-like responses, the AI community has been questioning whether this new technology had passed the <a href=\"https:\/\/en.wikipedia.org\/wiki\/Turing_test\" target=\"_blank\" rel=\"noopener\">Turing Test<\/a>.<\/em><\/p><p><em>Are we truly there? What happens behind the scenes, and will it disrupt delay analysis or the forensic world in general? In this 2-part introductory article, we will dive into the main techniques ChatGPT is based on. Then, as both a construction delay expert and CTO at Orizo Consult, I will provide insights into the potential use cases and future of the forensic industry.<\/em><\/p><p><em>One thing is for sure: if Microsoft is willing to <a href=\"https:\/\/www.theguardian.com\/technology\/2023\/jan\/23\/microsoft-confirms-multibillion-dollar-investment-in-firm-behind-chatgpt\" target=\"_blank\" rel=\"noopener\">invest 10 billion USD<\/a> into OpenAI, ChatGPT is no small beast.<\/em><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-d491b94 elementor-widget elementor-widget-spacer\" data-id=\"d491b94\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-46b367d elementor-widget elementor-widget-text-editor\" data-id=\"46b367d\" 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<h2>ChatGPT is powered by GPT-3<\/h2><p>ChatGPT is part of the GPT-3 family of Natural Language Processing (NLP) models. In order to understand ChatGPT, we will need to look at the foundations of GPT-3: a story of sentence statistics, semantical vectors, and word predictions.<\/p><h3>Generative Pre-trained Transformer<\/h3><p>The GPT acronym stands for Generative Pre-trained Transformer. It means the model creates new text, has been trained on non-specialized contents (it knows how to read), and relies upon the Transformer technology. A Transformer is an algorithm which is able to convert sentences into semantic meaning (their linguistic meaning) and then convert it back into a different sentence. The output text will depend on the task we wish to accomplish. A non-exhaustive list is presented below:<\/p><ul><li><strong>Rephrasing<\/strong> \u2013 Similar semantic meaning, similar length, different words.<\/li><li><strong>Shortening<\/strong> \u2013 Similar semantic meaning, shorter length.<\/li><li><strong>Extending<\/strong> \u2013 Similar semantic meaning, longer length.<\/li><li><strong>Translating<\/strong> \u2013 Similar semantic meaning, different dictionary of words.<\/li><li><strong>Extrapolating<\/strong> \u2013 Additional semantic meaning.<\/li><\/ul>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-7f4a0e6 elementor-widget elementor-widget-spacer\" data-id=\"7f4a0e6\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-795b2db elementor-widget elementor-widget-image\" data-id=\"795b2db\" 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<figure class=\"wp-caption\">\n\t\t\t\t\t\t\t\t\t\t<img loading=\"lazy\" decoding=\"async\" width=\"827\" height=\"541\" src=\"data:image\/svg+xml;charset=utf-8,%3Csvg xmlns%3D'http%3A%2F%2Fwww.w3.org%2F2000%2Fsvg' viewBox%3D'0 0 827 541'%2F%3E\" data-src=\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-1-Simplified-illustration-of-a-transformer.png\" class=\"lazy-load attachment-full size-full wp-image-6428\" alt=\"\" data-srcset=\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-1-Simplified-illustration-of-a-transformer.png 827w, https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-1-Simplified-illustration-of-a-transformer-300x196.png 300w, https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-1-Simplified-illustration-of-a-transformer-768x502.png 768w\" sizes=\"auto, (max-width: 827px) 100vw, 827px\" \/>\t\t\t\t\t\t\t\t\t\t\t<figcaption class=\"widget-image-caption wp-caption-text\">Figure 1: Simplified illustration of a transformer<\/figcaption>\n\t\t\t\t\t\t\t\t\t\t<\/figure>\n\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-dc92b78 elementor-widget elementor-widget-spacer\" data-id=\"dc92b78\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-d61cdac elementor-widget elementor-widget-text-editor\" data-id=\"d61cdac\" 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<h3 class=\"reader-text-block__heading2\">Infer semantic meaning from statistics<\/h3><p class=\"reader-text-block__paragraph\">During its training, GPT-3 read millions of documents (a\u00a0<em>corpus<\/em>) such as Wikipedia, books, news articles, product reviews, etc. From these readings, the model measured the frequency at which a word appeared close to other words in a sentence. That is, it measured the probability for a word to appear under a specific context of other words.<\/p><p class=\"reader-text-block__paragraph\">Let\u2019s consider a simple example where our corpus only includes two sentences: \u201chave a good day\u201d, and \u201chave a great day\u201d. The model would measure that \u2018good\u2019 and \u2018great\u2019 are equally used under the same context of words, i.e., at a 50% ratio.<\/p><p class=\"reader-text-block__paragraph\">The general idea behind this statistical analysis is that if two words are commonly used under similar contexts, their semantic meaning must be similar too. The model identifies the contexts under which words can be interchangeable to some degree, which then defines the strength of their semantic meaning in that context.<\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-7dc879c elementor-widget elementor-widget-spacer\" data-id=\"7dc879c\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-6b3daf0 elementor-widget elementor-widget-image\" data-id=\"6b3daf0\" 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<figure class=\"wp-caption\">\n\t\t\t\t\t\t\t\t\t\t<img loading=\"lazy\" decoding=\"async\" width=\"827\" height=\"484\" src=\"data:image\/svg+xml;charset=utf-8,%3Csvg xmlns%3D'http%3A%2F%2Fwww.w3.org%2F2000%2Fsvg' viewBox%3D'0 0 827 484'%2F%3E\" data-src=\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-2-Example-of-the-semantic-meaning-extraction-of-the-words-\u2018good-and-\u2018great-from-a-simple-corpus.png\" class=\"lazy-load attachment-full size-full wp-image-6430\" alt=\"\" data-srcset=\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-2-Example-of-the-semantic-meaning-extraction-of-the-words-\u2018good-and-\u2018great-from-a-simple-corpus.png 827w, https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-2-Example-of-the-semantic-meaning-extraction-of-the-words-\u2018good-and-\u2018great-from-a-simple-corpus-300x176.png 300w, https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-2-Example-of-the-semantic-meaning-extraction-of-the-words-\u2018good-and-\u2018great-from-a-simple-corpus-768x449.png 768w\" sizes=\"auto, (max-width: 827px) 100vw, 827px\" \/>\t\t\t\t\t\t\t\t\t\t\t<figcaption class=\"widget-image-caption wp-caption-text\">Figure 2: Example of the semantic meaning extraction of the words \u2018good\u2019 and \u2018great\u2019 from a simple corpus<\/figcaption>\n\t\t\t\t\t\t\t\t\t\t<\/figure>\n\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-048e148 elementor-widget elementor-widget-spacer\" data-id=\"048e148\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-4052aa6 elementor-widget elementor-widget-text-editor\" data-id=\"4052aa6\" 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<h2>Embeddings explained<\/h2><p>Embedding is the method through which the model stores semantic information in memory. We will see how the semantic meaning of a word can be quantified according to context, and how this can be extrapolated to capture the meaning of entire sentences or paragraphs.<\/p><h3>Embedding words into vectors<\/h3><p>Based on the statistical analysis mentioned in the previous section, the model is able to infer axes \u2013 also called <em>dimensions<\/em> \u2013 where the strength of a word for a given semantic meaning can be calculated.<\/p><p>Continuing with our example, we could imagine that \u2018good\u2019 and \u2018great\u2019 would both be captured to be strong in the \u2018positiveness\u2019 axis, where \u2018great\u2019 would probably be slightly stronger than \u2018good\u2019, and the other three words (\u2018have\u2019, \u2018a\u2019 and \u2018day\u2019) would have a neutral \u2018positiveness\u2019. Finally, if the words \u2018bad\u2019 or \u2018terrible\u2019 were also present in our corpus, we could expect that they would be characterized by weak and very weak \u2018positiveness\u2019. This strength is calculated as a percentage, which can then be stored in memory. A graphical representation of these values is shown below:<\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-6baa9cb elementor-widget elementor-widget-spacer\" data-id=\"6baa9cb\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-f7c524f elementor-widget elementor-widget-image\" data-id=\"f7c524f\" 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<figure class=\"wp-caption\">\n\t\t\t\t\t\t\t\t\t\t<img loading=\"lazy\" decoding=\"async\" width=\"827\" height=\"266\" src=\"data:image\/svg+xml;charset=utf-8,%3Csvg xmlns%3D'http%3A%2F%2Fwww.w3.org%2F2000%2Fsvg' viewBox%3D'0 0 827 266'%2F%3E\" data-src=\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-3-Locating-words-on-a-dimension-representing-\u2018positiveness.png\" class=\"lazy-load attachment-full size-full wp-image-6432\" alt=\"\" data-srcset=\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-3-Locating-words-on-a-dimension-representing-\u2018positiveness.png 827w, https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-3-Locating-words-on-a-dimension-representing-\u2018positiveness-300x96.png 300w, https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-3-Locating-words-on-a-dimension-representing-\u2018positiveness-768x247.png 768w\" sizes=\"auto, (max-width: 827px) 100vw, 827px\" \/>\t\t\t\t\t\t\t\t\t\t\t<figcaption class=\"widget-image-caption wp-caption-text\">Figure 3: Locating words on a dimension representing \u2018positiveness\u2019<\/figcaption>\n\t\t\t\t\t\t\t\t\t\t<\/figure>\n\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-ed01b8e elementor-widget elementor-widget-spacer\" data-id=\"ed01b8e\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-7b5af9f elementor-widget elementor-widget-text-editor\" data-id=\"7b5af9f\" 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<p>The more axes a model can dispose of, the better it gets at capturing the diversity of the meanings of a word. Thus, we may locate the words of our corpus against a second axis, which for example could reflect the size of things: \u2018largeness\u2019.<\/p><p>The two-axis description of these words can be visualized on the two-dimension chart below, where a few extra words have also been added, for the sake of further illustrating the principle.<\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-a0d8840 elementor-widget elementor-widget-spacer\" data-id=\"a0d8840\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-991ee34 elementor-widget elementor-widget-image\" data-id=\"991ee34\" 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<figure class=\"wp-caption\">\n\t\t\t\t\t\t\t\t\t\t<img loading=\"lazy\" decoding=\"async\" width=\"827\" height=\"477\" src=\"data:image\/svg+xml;charset=utf-8,%3Csvg xmlns%3D'http%3A%2F%2Fwww.w3.org%2F2000%2Fsvg' viewBox%3D'0 0 827 477'%2F%3E\" data-src=\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-4-Locating-words-on-a-2-dimensional-space-representing-\u2018positiveness-and-\u2018largeness.png\" class=\"lazy-load attachment-large size-large wp-image-6434\" alt=\"\" data-srcset=\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-4-Locating-words-on-a-2-dimensional-space-representing-\u2018positiveness-and-\u2018largeness.png 827w, https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-4-Locating-words-on-a-2-dimensional-space-representing-\u2018positiveness-and-\u2018largeness-300x173.png 300w, https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-4-Locating-words-on-a-2-dimensional-space-representing-\u2018positiveness-and-\u2018largeness-768x443.png 768w\" sizes=\"auto, (max-width: 827px) 100vw, 827px\" \/>\t\t\t\t\t\t\t\t\t\t\t<figcaption class=\"widget-image-caption wp-caption-text\">Figure 4: Locating words on a 2-dimensional space representing \u2018positiveness\u2019 and \u2018largeness\u2019<\/figcaption>\n\t\t\t\t\t\t\t\t\t\t<\/figure>\n\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-f552293 elementor-widget elementor-widget-spacer\" data-id=\"f552293\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-4dfe1d0 elementor-widget elementor-widget-text-editor\" data-id=\"4dfe1d0\" 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<p>We note that \u2018good\u2019 is neutral on the \u2018largeness\u2019 axis, whilst \u2018great\u2019 is strong on both the \u2018positiveness\u2019 and \u2018largeness\u2019 axes. Depending on the context of a sentence, the model will interpret the meaning of \u2018great\u2019 as either positive or large.<\/p><p>The semantic meaning of each word is stored in a 1-column table, where each row corresponds to the strength of the word in a given semantical dimension. These tables are called <em>embeddings<\/em>. Mathematically speaking, they are nothing more than multi-dimensional <em>vectors<\/em>.<\/p><p>In our example, we would store each word inside a 2-dimensional embedding. In reality though, the embeddings generated by GPT-3 include thousands of dimensions, each representing an axis of \u2018characteristic-ness\u2019.<\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-ace18d5 elementor-widget elementor-widget-spacer\" data-id=\"ace18d5\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-87bffb9 elementor-widget elementor-widget-image\" data-id=\"87bffb9\" 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<figure class=\"wp-caption\">\n\t\t\t\t\t\t\t\t\t\t<img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"401\" src=\"data:image\/svg+xml;charset=utf-8,%3Csvg xmlns%3D'http%3A%2F%2Fwww.w3.org%2F2000%2Fsvg' viewBox%3D'0 0 1024 401'%2F%3E\" data-src=\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-5-In-reality-GPT-3-embeddings-include-thousands-of-dimensions-1024x401.png\" class=\"lazy-load attachment-large size-large wp-image-6436\" alt=\"\" data-srcset=\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-5-In-reality-GPT-3-embeddings-include-thousands-of-dimensions-1024x401.png 1024w, https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-5-In-reality-GPT-3-embeddings-include-thousands-of-dimensions-300x118.png 300w, https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-5-In-reality-GPT-3-embeddings-include-thousands-of-dimensions-768x301.png 768w, https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-5-In-reality-GPT-3-embeddings-include-thousands-of-dimensions.png 1386w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/>\t\t\t\t\t\t\t\t\t\t\t<figcaption class=\"widget-image-caption wp-caption-text\">Figure 5: In reality, GPT-3 embeddings include thousands of dimensions<\/figcaption>\n\t\t\t\t\t\t\t\t\t\t<\/figure>\n\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-c886671 elementor-widget elementor-widget-spacer\" data-id=\"c886671\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-cea25d7 elementor-widget elementor-widget-text-editor\" data-id=\"cea25d7\" 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<p>As you may have already guessed, it becomes challenging for a human being to visualize that many dimensions. In practice, a semantical axis rarely reflects the strength of a word along a unique meaning. It rather reflects a trend resulting from the combination of multiple notions. As humans, we are mostly unable to decipher the full semantical notions behind each dimension. As a machine, GPT-3 is unable to understand the sense of the semantical notion it identified: it merely detected that specific words were used in a similar fashion under specific contexts.<\/p><h3>Adding or subtracting embeddings<\/h3><p>The magic of word embeddings starts with algebra. Since semantic meanings can be stored within vectors, which are nothing more than tables of numbers, why not try some maths? For example, how about summing the embedding of \u2018kid\u2019 with that of \u2018old\u2019, and hoping to get something like \u2018adult\u2019 as a result?<\/p><p>Back in 2013, Google researchers Tomas Mikolov et al. published their famous <a href=\"https:\/\/arxiv.org\/abs\/1301.3781\" target=\"_blank\" rel=\"noopener\">paper introducing word2vec<\/a>, a word embedding algorithm from which many subsequent embedding techniques were derived. At page 2, the paper says:<\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-884c807 elementor-widget elementor-widget-spacer\" data-id=\"884c807\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-d4c51cb elementor-widget elementor-widget-image\" data-id=\"d4c51cb\" 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<figure class=\"wp-caption\">\n\t\t\t\t\t\t\t\t\t\t<img loading=\"lazy\" decoding=\"async\" width=\"661\" height=\"83\" src=\"data:image\/svg+xml;charset=utf-8,%3Csvg xmlns%3D'http%3A%2F%2Fwww.w3.org%2F2000%2Fsvg' viewBox%3D'0 0 661 83'%2F%3E\" data-src=\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-6-Extract-from-Efficient-Estimation-of-Word-Representations-in-Vector-Space-arXiv-1301.3781-cs.CL_.png\" class=\"lazy-load attachment-large size-large wp-image-6438\" alt=\"\" data-srcset=\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-6-Extract-from-Efficient-Estimation-of-Word-Representations-in-Vector-Space-arXiv-1301.3781-cs.CL_.png 661w, https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-6-Extract-from-Efficient-Estimation-of-Word-Representations-in-Vector-Space-arXiv-1301.3781-cs.CL_-300x38.png 300w\" sizes=\"auto, (max-width: 661px) 100vw, 661px\" \/>\t\t\t\t\t\t\t\t\t\t\t<figcaption class=\"widget-image-caption wp-caption-text\">Figure 6: Extract from \u201cEfficient Estimation of Word Representations in Vector Space\u201d, arXiv:1301.3781 [cs.CL]<\/figcaption>\n\t\t\t\t\t\t\t\t\t\t<\/figure>\n\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-b238ee9 elementor-widget elementor-widget-spacer\" data-id=\"b238ee9\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-0dca387 elementor-widget elementor-widget-text-editor\" data-id=\"0dca387\" 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<p>The word2vec algorithm paved the road toward the era of semantical algebra. It was now possible to infer words from other words, based on their linguistical meaning.<\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-4a0559d elementor-widget elementor-widget-spacer\" data-id=\"4a0559d\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-9c74ad2 elementor-widget elementor-widget-image\" data-id=\"9c74ad2\" 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<figure class=\"wp-caption\">\n\t\t\t\t\t\t\t\t\t\t<img loading=\"lazy\" decoding=\"async\" width=\"590\" height=\"163\" src=\"data:image\/svg+xml;charset=utf-8,%3Csvg xmlns%3D'http%3A%2F%2Fwww.w3.org%2F2000%2Fsvg' viewBox%3D'0 0 590 163'%2F%3E\" data-src=\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-7-Adding-and-subtracting-embeddings.png\" class=\"lazy-load attachment-full size-full wp-image-6440\" alt=\"\" data-srcset=\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-7-Adding-and-subtracting-embeddings.png 590w, https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-7-Adding-and-subtracting-embeddings-300x83.png 300w\" sizes=\"auto, (max-width: 590px) 100vw, 590px\" \/>\t\t\t\t\t\t\t\t\t\t\t<figcaption class=\"widget-image-caption wp-caption-text\">Figure 7: Adding and subtracting embeddings<\/figcaption>\n\t\t\t\t\t\t\t\t\t\t<\/figure>\n\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-573691c elementor-widget elementor-widget-spacer\" data-id=\"573691c\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-97610e1 elementor-widget elementor-widget-text-editor\" data-id=\"97610e1\" 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<p>Behind the scenes, what the model identified was that the difference between not \u2018old\u2019 and \u2018old\u2019 was similar to the difference between \u2018adult\u2019 and \u2018kid\u2019. Likewise, it identified that the main difference between \u2018man\u2019 and \u2018woman\u2019 was similar to the main difference between \u2018king\u2019 and \u2018queen\u2019. Considering a 2D representation along the relevant semantical dimensions, the additions and subtractions would show as follows:<\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-314a77c elementor-widget elementor-widget-spacer\" data-id=\"314a77c\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-97d6a60 elementor-widget elementor-widget-image\" data-id=\"97d6a60\" 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<figure class=\"wp-caption\">\n\t\t\t\t\t\t\t\t\t\t<img loading=\"lazy\" decoding=\"async\" width=\"827\" height=\"313\" src=\"data:image\/svg+xml;charset=utf-8,%3Csvg xmlns%3D'http%3A%2F%2Fwww.w3.org%2F2000%2Fsvg' viewBox%3D'0 0 827 313'%2F%3E\" data-src=\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-8-Inferring-words-by-adding-or-subtracting-other-words.png\" class=\"lazy-load attachment-large size-large wp-image-6442\" alt=\"\" data-srcset=\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-8-Inferring-words-by-adding-or-subtracting-other-words.png 827w, https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-8-Inferring-words-by-adding-or-subtracting-other-words-300x114.png 300w, https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-8-Inferring-words-by-adding-or-subtracting-other-words-768x291.png 768w\" sizes=\"auto, (max-width: 827px) 100vw, 827px\" \/>\t\t\t\t\t\t\t\t\t\t\t<figcaption class=\"widget-image-caption wp-caption-text\">Figure 8: Inferring words by adding or subtracting other words<\/figcaption>\n\t\t\t\t\t\t\t\t\t\t<\/figure>\n\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-f0779e8 elementor-widget elementor-widget-spacer\" data-id=\"f0779e8\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-e0879fd elementor-widget elementor-widget-text-editor\" data-id=\"e0879fd\" 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<h3>Paragraph embeddings<\/h3><p>Just like adding embeddings allows one to infer new words, calculating the average of a sub-set of embeddings allows one to derive the overall meaning of a group of words. For example, one may derive the meaning of a sentence by averaging the embeddings of all the words the sentence is made of. One of the novelties of the Transformer technology used by GPT-3 is that it considers the position of the words in the sentence, hence capturing additional semantic information when compared to grossly calculating an average.<\/p><p>Once able to calculate the embedding of a sentence, the same principle can be applied in order to calculate the embeddings of paragraphs.<\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-b631212 elementor-widget elementor-widget-spacer\" data-id=\"b631212\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-43b5b4b elementor-widget elementor-widget-text-editor\" data-id=\"43b5b4b\" 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<h2>GPT-3 leverages embeddings to re-formulate text<\/h2><p>We have seen that one of the fundamental techniques behind GPT-3 is the embedding of words, sentences, or paragraphs. In the second part of this article, we will see how the model is able to translate embeddings back into sentences that are structurally and semantically correct.<\/p><p>Before heading there, we can however already get a first feel of what GPT-3 and by extension ChatGPT may be good at:<\/p><ul><li>Re-writing a paragraph with a similar meaning but different words.<\/li><li>Generating summaries: re-write with fewer words.<\/li><li>Converting a paragraph into keywords.<\/li><li>Performing rich searches by meaning as opposed to keywords.<\/li><\/ul><p>A few use-case examples are explored below.<\/p><h3>Paraphrasing<\/h3><p>Probably one of the most direct techniques used with ChatGPT, paraphrasing allows one to re-write a paragraph whilst maintaining its overall meaning. It is also possible to train the model to amplify certain aspects of the rewording.<\/p><p>In the example below, we ask ChatGPT to rewrite into British English, whilst at the same time aiming for a specific audience and requesting to improve readability.<\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-22d196e elementor-widget elementor-widget-spacer\" data-id=\"22d196e\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-85e2d21 elementor-widget elementor-widget-image\" data-id=\"85e2d21\" 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<figure class=\"wp-caption\">\n\t\t\t\t\t\t\t\t\t\t<img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"348\" src=\"data:image\/svg+xml;charset=utf-8,%3Csvg xmlns%3D'http%3A%2F%2Fwww.w3.org%2F2000%2Fsvg' viewBox%3D'0 0 1024 348'%2F%3E\" data-src=\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-9-Asking-ChatGPT-to-improve-a-poorly-written-paragraph-1024x348.png\" class=\"lazy-load attachment-large size-large wp-image-6444\" alt=\"\" data-srcset=\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-9-Asking-ChatGPT-to-improve-a-poorly-written-paragraph-1024x348.png 1024w, https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-9-Asking-ChatGPT-to-improve-a-poorly-written-paragraph-300x102.png 300w, https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-9-Asking-ChatGPT-to-improve-a-poorly-written-paragraph-768x261.png 768w, https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-9-Asking-ChatGPT-to-improve-a-poorly-written-paragraph.png 1342w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/>\t\t\t\t\t\t\t\t\t\t\t<figcaption class=\"widget-image-caption wp-caption-text\">Figure 9: Asking ChatGPT to improve a poorly written paragraph<\/figcaption>\n\t\t\t\t\t\t\t\t\t\t<\/figure>\n\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-79b926c elementor-widget elementor-widget-spacer\" data-id=\"79b926c\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-e9a1d63 elementor-widget elementor-widget-text-editor\" data-id=\"e9a1d63\" 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<h3>Summarising<\/h3><p>Another technique often leveraged is GPT-3\u2019s ability to rewrite a paragraph or set of paragraphs with fewer words, whilst preventing the loss of the most relevant meaning.<\/p><p>This is a powerful feature that can enable senior staff to review documents they may otherwise never get the time to open and would typically delegate to junior members. One should nonetheless not negate the risk that GPT miscalculate the relevance of some information in the original text, which would then be omitted in the summary. There is always a risk in delegating. Is it riskier to rely on AI instead of junior staff? I guess the answer to this question will evolve dramatically in the years to come.<\/p><p>One should always be cautious not to delegate the screening of potentially critical documents. For the rest, the right balance resides in the odd equilibrium between time, money, and the magnitude of the assessment. AI would probably allow screening documents no one would have ever considered before.<\/p><p>In an era where experts and lawyers are handed over hundreds of gigabits of documents, whilst at the same time pressured for a cost-effective delivery, AI-assisted screening is an advantage. I also suspect it will enable the claiming of damages which up to now were often too costly for being properly substantiated. Would you consider preparing a fully-fledged disruption claim?<\/p><h3>Keywords extraction<\/h3><p>This technique, which consists in summarizing a paragraph into just a few words, is based on the model\u2019s ability to identify the relevance or centrality of each word to the overall paragraph: the <em>salience<\/em>. The most salient words can then be combined in order to calculate one or several keywords which best reflect their meaning.<\/p><p>The technique can be used with emails, reports, claims, or any other business documents to generate labels. These documents may then be grouped or searched by label. It is also possible to train the model to return the most relevant keyword amongst a given list, or to provide their relevance. Whilst it is not the most ergonomic user interface for performing the task, ChatGPT is able to do it:<\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-d426212 elementor-widget elementor-widget-spacer\" data-id=\"d426212\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-2053e49 elementor-widget elementor-widget-image\" data-id=\"2053e49\" 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<figure class=\"wp-caption\">\n\t\t\t\t\t\t\t\t\t\t<img loading=\"lazy\" decoding=\"async\" width=\"661\" height=\"486\" src=\"data:image\/svg+xml;charset=utf-8,%3Csvg xmlns%3D'http%3A%2F%2Fwww.w3.org%2F2000%2Fsvg' viewBox%3D'0 0 661 486'%2F%3E\" data-src=\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-10-Asking-ChatGPT-to-provide-the-relevance-of-a-set-of-keywords-in-a-given-paragraph.png\" class=\"lazy-load attachment-large size-large wp-image-6446\" alt=\"\" data-srcset=\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-10-Asking-ChatGPT-to-provide-the-relevance-of-a-set-of-keywords-in-a-given-paragraph.png 661w, https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-10-Asking-ChatGPT-to-provide-the-relevance-of-a-set-of-keywords-in-a-given-paragraph-300x221.png 300w\" sizes=\"auto, (max-width: 661px) 100vw, 661px\" \/>\t\t\t\t\t\t\t\t\t\t\t<figcaption class=\"widget-image-caption wp-caption-text\">Figure 10: Asking ChatGPT to provide the relevance of a set of keywords in a given paragraph<\/figcaption>\n\t\t\t\t\t\t\t\t\t\t<\/figure>\n\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-d10752f elementor-widget elementor-widget-spacer\" data-id=\"d10752f\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-ab227a2 elementor-widget elementor-widget-text-editor\" data-id=\"ab227a2\" 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<p>During document screening, one may wish to identify which sections of which documents are the most relevant to specific matters. These sections could then be grouped accordingly.<\/p><p>In the future, we could also envisage a similar system where electronic Discovery would be governed by such keyword relevance queries.<\/p><h3>Semantic searches<\/h3><p>Since the inception of web search engines, we have been used to search by keyword. This technique is characterized by two main flaws: (i) a single keyword may contain multiple meanings, which adds noise to the search result, and (ii) an entry containing synonyms of the searched keyword may not show in the results.<\/p><p>There have been improvements with search engines, which are notably now able to deal with synonyms, but we are still far from being able to search a precise idea. What is more, it is also hard to add context in order to refine a search. ChatGPT allows to express an idea and search for it in the corpus.<\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-389349c elementor-widget elementor-widget-spacer\" data-id=\"389349c\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-15c6a88 elementor-widget elementor-widget-image\" data-id=\"15c6a88\" 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<figure class=\"wp-caption\">\n\t\t\t\t\t\t\t\t\t\t<img loading=\"lazy\" decoding=\"async\" width=\"661\" height=\"429\" src=\"data:image\/svg+xml;charset=utf-8,%3Csvg xmlns%3D'http%3A%2F%2Fwww.w3.org%2F2000%2Fsvg' viewBox%3D'0 0 661 429'%2F%3E\" data-src=\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-11-Asking-ChatGPT-about-the-enforceability-of-a-contract-clause-under-a-specific-context.png\" class=\"lazy-load attachment-large size-large wp-image-6448\" alt=\"\" data-srcset=\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-11-Asking-ChatGPT-about-the-enforceability-of-a-contract-clause-under-a-specific-context.png 661w, https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/Fig-11-Asking-ChatGPT-about-the-enforceability-of-a-contract-clause-under-a-specific-context-300x195.png 300w\" sizes=\"auto, (max-width: 661px) 100vw, 661px\" \/>\t\t\t\t\t\t\t\t\t\t\t<figcaption class=\"widget-image-caption wp-caption-text\">Figure 11: Asking ChatGPT about the enforceability of a contract clause under a specific context<\/figcaption>\n\t\t\t\t\t\t\t\t\t\t<\/figure>\n\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-e967923 elementor-widget elementor-widget-spacer\" data-id=\"e967923\" 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<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-db7cc49 elementor-widget elementor-widget-text-editor\" data-id=\"db7cc49\" 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<h3 class=\"reader-text-block__heading2\"><em>Coming soon&#8230;<\/em><\/h3><p class=\"reader-text-block__paragraph\">In the second part of this article, we will see how sentence completion works, which is the principle that allows ChatGPT to convert embeddings back into human-like responses. We will explore the two techniques commonly used to train natural language AI to respond accurately, and will see how this could offer opportunities for the forensic industry.<\/p>\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\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t\t\t\t<\/div>\n\t\t","protected":false},"excerpt":{"rendered":"<p>Explaining the mechanics behind ChatGPT and the future prospects for our industry. This article will be delivered in two parts, each focusing on a specific technology ChatGPT is based on, followed by some concrete use cases.<\/p>\n","protected":false},"author":2,"featured_media":6457,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_monsterinsights_skip_tracking":false,"_monsterinsights_sitenote_active":false,"_monsterinsights_sitenote_note":"","_monsterinsights_sitenote_category":0,"footnotes":""},"categories":[3],"tags":[72,90,70,71,73],"class_list":["post-6320","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-insights","tag-construction","tag-cutting-edge-delay-analysis-series","tag-delay-analysis","tag-forensics","tag-project-management"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v21.1 - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>Cutting Edge Delay Analysis Series - ChatGPT technologies: a disruption to the forensic industry? \u2013 Part 1: Embeddings | Orizo Consult International<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.orizoconsult.com\/fr\/insights\/chatgpt-technologies-disruption-forensic-industry-part-1-embeddings\/\" \/>\n<meta property=\"og:locale\" content=\"fr_FR\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Cutting Edge Delay Analysis Series - ChatGPT technologies: a disruption to the forensic industry? \u2013 Part 1: Embeddings | Orizo Consult International\" \/>\n<meta property=\"og:description\" content=\"Explaining the mechanics behind ChatGPT and the future prospects for our industry. This article will be delivered in two parts, each focusing on a specific technology ChatGPT is based on, followed by some concrete use cases.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.orizoconsult.com\/insights\/chatgpt-technologies-disruption-forensic-industry-part-1-embeddings\/\" \/>\n<meta property=\"og:site_name\" content=\"Orizo Consult International\" \/>\n<meta property=\"article:published_time\" content=\"2023-02-06T08:00:28+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2023-04-27T15:28:50+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/pexels-pixabay-267669-scaled.jpg\" \/>\n\t<meta property=\"og:image:width\" content=\"2560\" \/>\n\t<meta property=\"og:image:height\" content=\"1706\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/jpeg\" \/>\n<meta name=\"author\" content=\"Ama\u00ebl Olivier\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u00c9crit par\" \/>\n\t<meta name=\"twitter:data1\" content=\"Ama\u00ebl Olivier\" \/>\n\t<meta name=\"twitter:label2\" content=\"Dur\u00e9e de lecture estim\u00e9e\" \/>\n\t<meta name=\"twitter:data2\" content=\"10 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"ScholarlyArticle\",\"@id\":\"https:\/\/www.orizoconsult.com\/insights\/chatgpt-technologies-disruption-forensic-industry-part-1-embeddings\/#article\",\"isPartOf\":{\"@id\":\"https:\/\/www.orizoconsult.com\/insights\/chatgpt-technologies-disruption-forensic-industry-part-1-embeddings\/\"},\"author\":{\"name\":\"Ama\u00ebl Olivier\",\"@id\":\"https:\/\/www.orizoconsult.com\/fr\/#\/schema\/person\/3effe2a0a4253a67ad3b84b82e57a708\"},\"headline\":\"Cutting Edge Delay Analysis Series &#8211; ChatGPT technologies: a disruption to the forensic industry? \u2013 Part 1: Embeddings\",\"datePublished\":\"2023-02-06T08:00:28+00:00\",\"dateModified\":\"2023-04-27T15:28:50+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\/\/www.orizoconsult.com\/insights\/chatgpt-technologies-disruption-forensic-industry-part-1-embeddings\/\"},\"wordCount\":2080,\"publisher\":{\"@id\":\"https:\/\/www.orizoconsult.com\/fr\/#organization\"},\"keywords\":[\"construction\",\"Cutting Edge Delay Analysis Series\",\"delay-analysis\",\"forensics\",\"project-management\"],\"articleSection\":[\"Industry Insights\"],\"inLanguage\":\"fr-FR\"},{\"@type\":\"WebPage\",\"@id\":\"https:\/\/www.orizoconsult.com\/insights\/chatgpt-technologies-disruption-forensic-industry-part-1-embeddings\/\",\"url\":\"https:\/\/www.orizoconsult.com\/insights\/chatgpt-technologies-disruption-forensic-industry-part-1-embeddings\/\",\"name\":\"Cutting Edge Delay Analysis Series - ChatGPT technologies: a disruption to the forensic industry? \u2013 Part 1: Embeddings | Orizo Consult International\",\"isPartOf\":{\"@id\":\"https:\/\/www.orizoconsult.com\/fr\/#website\"},\"datePublished\":\"2023-02-06T08:00:28+00:00\",\"dateModified\":\"2023-04-27T15:28:50+00:00\",\"breadcrumb\":{\"@id\":\"https:\/\/www.orizoconsult.com\/insights\/chatgpt-technologies-disruption-forensic-industry-part-1-embeddings\/#breadcrumb\"},\"inLanguage\":\"fr-FR\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/www.orizoconsult.com\/insights\/chatgpt-technologies-disruption-forensic-industry-part-1-embeddings\/\"]}]},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/www.orizoconsult.com\/insights\/chatgpt-technologies-disruption-forensic-industry-part-1-embeddings\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\/\/www.orizoconsult.com\/fr\/fr-home\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Cutting Edge Delay Analysis Series &#8211; ChatGPT technologies: a disruption to the forensic industry? \u2013 Part 1: Embeddings\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/www.orizoconsult.com\/fr\/#website\",\"url\":\"https:\/\/www.orizoconsult.com\/fr\/\",\"name\":\"Orizo Consult International\",\"description\":\"Cutting Edge Expert Services\",\"publisher\":{\"@id\":\"https:\/\/www.orizoconsult.com\/fr\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/www.orizoconsult.com\/fr\/?s={search_term_string}\"},\"query-input\":\"required name=search_term_string\"}],\"inLanguage\":\"fr-FR\"},{\"@type\":\"Organization\",\"@id\":\"https:\/\/www.orizoconsult.com\/fr\/#organization\",\"name\":\"Orizo Consult\",\"url\":\"https:\/\/www.orizoconsult.com\/fr\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"fr-FR\",\"@id\":\"https:\/\/www.orizoconsult.com\/fr\/#\/schema\/logo\/image\/\",\"url\":\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2021\/07\/Main-Horizontal-2.png\",\"contentUrl\":\"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2021\/07\/Main-Horizontal-2.png\",\"width\":300,\"height\":105,\"caption\":\"Orizo Consult\"},\"image\":{\"@id\":\"https:\/\/www.orizoconsult.com\/fr\/#\/schema\/logo\/image\/\"},\"sameAs\":[\"https:\/\/es.linkedin.com\/company\/orizo-consult\"]},{\"@type\":\"Person\",\"@id\":\"https:\/\/www.orizoconsult.com\/fr\/#\/schema\/person\/3effe2a0a4253a67ad3b84b82e57a708\",\"name\":\"Ama\u00ebl Olivier\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"fr-FR\",\"@id\":\"https:\/\/www.orizoconsult.com\/fr\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/secure.gravatar.com\/avatar\/a5d3975779ab3847fa2a1a184e4c76c3bcc7e51a4f1d397c8896d49cbe7db42e?s=96&d=mm&r=g\",\"contentUrl\":\"https:\/\/secure.gravatar.com\/avatar\/a5d3975779ab3847fa2a1a184e4c76c3bcc7e51a4f1d397c8896d49cbe7db42e?s=96&d=mm&r=g\",\"caption\":\"Ama\u00ebl Olivier\"},\"description\":\"Construction delay expert Ama\u00ebl Olivier is the Founding partner of Orizo Consult International. He cumulates over a decade of delay analysis know-how, including the review of major landmark projects across more than 25 countries. He excels in complex infrastructure and energy investments, having assessed projects ranging from 10 to 3,500 million USD in capital value. He testifies in English, French, and Spanish. The concepts presented in this article are orientative and have been simplified for the purpose of addressing a wide audience. The opinions are those of the author at the time of writing, not necessarily those of Orizo Consult International. Each case includes its own particularities, which is why we do not pretend to be providing professional advice through this content.\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"Cutting Edge Delay Analysis Series - ChatGPT technologies: a disruption to the forensic industry? \u2013 Part 1: Embeddings | Orizo Consult International","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:\/\/www.orizoconsult.com\/fr\/insights\/chatgpt-technologies-disruption-forensic-industry-part-1-embeddings\/","og_locale":"fr_FR","og_type":"article","og_title":"Cutting Edge Delay Analysis Series - ChatGPT technologies: a disruption to the forensic industry? \u2013 Part 1: Embeddings | Orizo Consult International","og_description":"Explaining the mechanics behind ChatGPT and the future prospects for our industry. This article will be delivered in two parts, each focusing on a specific technology ChatGPT is based on, followed by some concrete use cases.","og_url":"https:\/\/www.orizoconsult.com\/insights\/chatgpt-technologies-disruption-forensic-industry-part-1-embeddings\/","og_site_name":"Orizo Consult International","article_published_time":"2023-02-06T08:00:28+00:00","article_modified_time":"2023-04-27T15:28:50+00:00","og_image":[{"width":2560,"height":1706,"url":"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2023\/02\/pexels-pixabay-267669-scaled.jpg","type":"image\/jpeg"}],"author":"Ama\u00ebl Olivier","twitter_card":"summary_large_image","twitter_misc":{"\u00c9crit par":"Ama\u00ebl Olivier","Dur\u00e9e de lecture estim\u00e9e":"10 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"ScholarlyArticle","@id":"https:\/\/www.orizoconsult.com\/insights\/chatgpt-technologies-disruption-forensic-industry-part-1-embeddings\/#article","isPartOf":{"@id":"https:\/\/www.orizoconsult.com\/insights\/chatgpt-technologies-disruption-forensic-industry-part-1-embeddings\/"},"author":{"name":"Ama\u00ebl Olivier","@id":"https:\/\/www.orizoconsult.com\/fr\/#\/schema\/person\/3effe2a0a4253a67ad3b84b82e57a708"},"headline":"Cutting Edge Delay Analysis Series &#8211; ChatGPT technologies: a disruption to the forensic industry? \u2013 Part 1: Embeddings","datePublished":"2023-02-06T08:00:28+00:00","dateModified":"2023-04-27T15:28:50+00:00","mainEntityOfPage":{"@id":"https:\/\/www.orizoconsult.com\/insights\/chatgpt-technologies-disruption-forensic-industry-part-1-embeddings\/"},"wordCount":2080,"publisher":{"@id":"https:\/\/www.orizoconsult.com\/fr\/#organization"},"keywords":["construction","Cutting Edge Delay Analysis Series","delay-analysis","forensics","project-management"],"articleSection":["Industry Insights"],"inLanguage":"fr-FR"},{"@type":"WebPage","@id":"https:\/\/www.orizoconsult.com\/insights\/chatgpt-technologies-disruption-forensic-industry-part-1-embeddings\/","url":"https:\/\/www.orizoconsult.com\/insights\/chatgpt-technologies-disruption-forensic-industry-part-1-embeddings\/","name":"Cutting Edge Delay Analysis Series - ChatGPT technologies: a disruption to the forensic industry? \u2013 Part 1: Embeddings | Orizo Consult International","isPartOf":{"@id":"https:\/\/www.orizoconsult.com\/fr\/#website"},"datePublished":"2023-02-06T08:00:28+00:00","dateModified":"2023-04-27T15:28:50+00:00","breadcrumb":{"@id":"https:\/\/www.orizoconsult.com\/insights\/chatgpt-technologies-disruption-forensic-industry-part-1-embeddings\/#breadcrumb"},"inLanguage":"fr-FR","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.orizoconsult.com\/insights\/chatgpt-technologies-disruption-forensic-industry-part-1-embeddings\/"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/www.orizoconsult.com\/insights\/chatgpt-technologies-disruption-forensic-industry-part-1-embeddings\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/www.orizoconsult.com\/fr\/fr-home\/"},{"@type":"ListItem","position":2,"name":"Cutting Edge Delay Analysis Series &#8211; ChatGPT technologies: a disruption to the forensic industry? \u2013 Part 1: Embeddings"}]},{"@type":"WebSite","@id":"https:\/\/www.orizoconsult.com\/fr\/#website","url":"https:\/\/www.orizoconsult.com\/fr\/","name":"Orizo Consult International","description":"Cutting Edge Expert Services","publisher":{"@id":"https:\/\/www.orizoconsult.com\/fr\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.orizoconsult.com\/fr\/?s={search_term_string}"},"query-input":"required name=search_term_string"}],"inLanguage":"fr-FR"},{"@type":"Organization","@id":"https:\/\/www.orizoconsult.com\/fr\/#organization","name":"Orizo Consult","url":"https:\/\/www.orizoconsult.com\/fr\/","logo":{"@type":"ImageObject","inLanguage":"fr-FR","@id":"https:\/\/www.orizoconsult.com\/fr\/#\/schema\/logo\/image\/","url":"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2021\/07\/Main-Horizontal-2.png","contentUrl":"https:\/\/www.orizoconsult.com\/wp-content\/uploads\/2021\/07\/Main-Horizontal-2.png","width":300,"height":105,"caption":"Orizo Consult"},"image":{"@id":"https:\/\/www.orizoconsult.com\/fr\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/es.linkedin.com\/company\/orizo-consult"]},{"@type":"Person","@id":"https:\/\/www.orizoconsult.com\/fr\/#\/schema\/person\/3effe2a0a4253a67ad3b84b82e57a708","name":"Ama\u00ebl Olivier","image":{"@type":"ImageObject","inLanguage":"fr-FR","@id":"https:\/\/www.orizoconsult.com\/fr\/#\/schema\/person\/image\/","url":"https:\/\/secure.gravatar.com\/avatar\/a5d3975779ab3847fa2a1a184e4c76c3bcc7e51a4f1d397c8896d49cbe7db42e?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/a5d3975779ab3847fa2a1a184e4c76c3bcc7e51a4f1d397c8896d49cbe7db42e?s=96&d=mm&r=g","caption":"Ama\u00ebl Olivier"},"description":"Construction delay expert Ama\u00ebl Olivier is the Founding partner of Orizo Consult International. He cumulates over a decade of delay analysis know-how, including the review of major landmark projects across more than 25 countries. He excels in complex infrastructure and energy investments, having assessed projects ranging from 10 to 3,500 million USD in capital value. He testifies in English, French, and Spanish. The concepts presented in this article are orientative and have been simplified for the purpose of addressing a wide audience. The opinions are those of the author at the time of writing, not necessarily those of Orizo Consult International. Each case includes its own particularities, which is why we do not pretend to be providing professional advice through this content."}]}},"_links":{"self":[{"href":"https:\/\/www.orizoconsult.com\/fr\/wp-json\/wp\/v2\/posts\/6320","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.orizoconsult.com\/fr\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.orizoconsult.com\/fr\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.orizoconsult.com\/fr\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.orizoconsult.com\/fr\/wp-json\/wp\/v2\/comments?post=6320"}],"version-history":[{"count":29,"href":"https:\/\/www.orizoconsult.com\/fr\/wp-json\/wp\/v2\/posts\/6320\/revisions"}],"predecessor-version":[{"id":6943,"href":"https:\/\/www.orizoconsult.com\/fr\/wp-json\/wp\/v2\/posts\/6320\/revisions\/6943"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.orizoconsult.com\/fr\/wp-json\/wp\/v2\/media\/6457"}],"wp:attachment":[{"href":"https:\/\/www.orizoconsult.com\/fr\/wp-json\/wp\/v2\/media?parent=6320"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.orizoconsult.com\/fr\/wp-json\/wp\/v2\/categories?post=6320"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.orizoconsult.com\/fr\/wp-json\/wp\/v2\/tags?post=6320"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}