{"id":70838,"date":"2022-12-01T13:34:54","date_gmt":"2022-12-01T11:34:54","guid":{"rendered":"https:\/\/forklog.com\/en\/?p=70838"},"modified":"2025-09-08T04:32:29","modified_gmt":"2025-09-08T01:32:29","slug":"openai-unveils-chatgpt-a-chatbot-designed-for-dialogue","status":"publish","type":"post","link":"https:\/\/u1f987.com\/en\/openai-unveils-chatgpt-a-chatbot-designed-for-dialogue\/","title":{"rendered":"OpenAI unveils ChatGPT, a chatbot designed for dialogue"},"content":{"rendered":"<p>OpenAI released the ChatGPT chatbot, which can answer questions, acknowledge mistakes, argue and reject inappropriate requests.<\/p>\n<blockquote class=\"twitter-tweet\">\n<p lang=\"en\" dir=\"ltr\">Try talking with ChatGPT, our new AI system which is optimized for dialogue. Your feedback will help us improve it. <a href=\"https:\/\/t.co\/sHDm57g3Kr\">https:\/\/t.co\/sHDm57g3Kr<\/a><\/p>\n<p>\u2014 OpenAI (@OpenAI) <a href=\"https:\/\/twitter.com\/OpenAI\/status\/1598014522098208769?ref_src=twsrc%5Etfw\">November 30, 2022<\/a><\/p><\/blockquote>\n<p> <script async src=\"https:\/\/platform.twitter.com\/widgets.js\" charset=\"utf-8\"><\/script><\/p>\n<p>The developers built the model using reinforcement learning from human feedback (RLHF). They used the same methods as InstructGPT, but with additional data from human dialogues.<\/p>\n<p>To assemble the conversation dataset, the developers enlisted so-called trainers. They conducted conversations on behalf of a human and an AI assistant. Trainers also had access to model-generated proposals that helped them write answers.<\/p>\n<p>For the reinforcement learning reward model, the team recorded conversations between the trainer and the chatbot. Then it randomly selected AI-generated responses and asked the instructors to rank them.<\/p>\n<p>To improve the model&#8217;s accuracy, the developers employed proximal policy optimization. They performed several iterations of this process.<\/p>\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"608\" src=\"https:\/\/u1f987.com\/wp-content\/uploads\/Chatgpt_diagram-1024x608.png\" alt=\"OpenAI unveils ChatGPT chatbot\" class=\"wp-image-192183\" srcset=\"https:\/\/u1f987.com\/wp-content\/uploads\/Chatgpt_diagram-1024x608.png 1024w, https:\/\/u1f987.com\/wp-content\/uploads\/Chatgpt_diagram-300x178.png 300w, https:\/\/u1f987.com\/wp-content\/uploads\/Chatgpt_diagram-768x456.png 768w, https:\/\/u1f987.com\/wp-content\/uploads\/Chatgpt_diagram.png 1140w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption>Process of training ChatGPT. Data: OpenAI.<\/figcaption><\/figure>\n<p>According to the developers, the model has a number of limitations. Despite plausible responses, ChatGPT often makes mistakes, is sensitive to wording, overuses certain phrases, demonstrates bias and infers context.<\/p>\n<p>However, the company said that the experience deploying GPT-3 and Codex helped increase the model&#8217;s safety and reduce the number of unreliable answers thanks to RLHF.<\/p>\n<p>ChatGPT <a href=\"https:\/\/chat.openai.com\/\" target=\"_blank\" rel=\"noopener nofollow\" title=\"\">available<\/a> for free. The developers asked users to provide feedback on problematic outputs from the model, false triggers of the external content filter, which is part of the interface.<\/p>\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>&#8220;We are particularly interested in feedback [&#8230;], which helps us identify and understand new risks and possible measures to address them,&#8221; the press release says.<\/p>\n<\/blockquote>\n<p>According to the developers, the information gathered will assist in developing more modern and advanced language models.<\/p>\n<p>ChatGPT was trained on the Azure AI supercomputer. The chatbot is a refinement of the GPT-3.5 model, whose development was completed in early 2022.<\/p>\n<p>In September, OpenAI unveiled <a href=\"https:\/\/u1f987.com\/en\/news\/openai-unveils-whisper-an-open-source-speech-recognition-system\">speech-recognition system Whisper<\/a>, enabling transcription in multiple languages.<\/p>\n<p>In January the company released <a href=\"https:\/\/u1f987.com\/en\/news\/openai-has-created-a-less-toxic-version-of-gpt-3\">a new version of GPT-3<\/a>, which produces fewer offensive expressions, disinformation and errors overall.<\/p>\n<p>Subscribe to ForkLog AI on Telegram: <a href=\"https:\/\/t.me\/forklogAI\" target=\"_blank\" rel=\"noopener nofollow\" title=\"\">ForkLog AI<\/a> \u2014 all the news from the world of AI!<\/p>\n","protected":false},"excerpt":{"rendered":"<p>OpenAI released the ChatGPT chatbot, which can answer questions, acknowledge mistakes, argue and reject inappropriate requests.<\/p>\n","protected":false},"author":1,"featured_media":70839,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"select":"1","news_style_id":"1","cryptorium_level":"","_short_excerpt_text":"","creation_source":"","_metatest_mainpost_news_update":false,"footnotes":""},"categories":[3],"tags":[438,1190],"class_list":["post-70838","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-news-and-analysis","tag-artificial-intelligence","tag-openai"],"aioseo_notices":[],"amp_enabled":true,"views":"54","promo_type":"1","layout_type":"1","short_excerpt":"","is_update":"","_links":{"self":[{"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/posts\/70838","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/comments?post=70838"}],"version-history":[{"count":1,"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/posts\/70838\/revisions"}],"predecessor-version":[{"id":70840,"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/posts\/70838\/revisions\/70840"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/media\/70839"}],"wp:attachment":[{"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/media?parent=70838"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/categories?post=70838"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/tags?post=70838"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}