\n<\/p>\n
In response to pressure from rivals including Chinese AI company DeepSeek<\/a>, OpenAI is changing the way its newest AI model, o3-mini<\/a>, communicates its step-by-step \u201cthought\u201d process.<\/p>\n On Thursday, OpenAI announced<\/a> that free and paid users of ChatGPT, the company\u2019s AI-powered chatbot platform, will see an updated \u201cchain of thought\u201d that shows more of the model\u2019s \u201creasoning\u201d steps and how it arrived at answers to questions. Subscribers to premium ChatGPT plans who use o3-mini in the \u201chigh reasoning\u201d configuration will also see this updated readout, according to OpenAI.<\/p>\n \u201cWe\u2019re introducing an updated [chain of thought] for o3-mini designed to make it easier for people to understand how the model thinks,\u201d an OpenAI spokesperson told TechCruch via email. \u201cWith this update, you will be able to follow the model\u2019s reasoning, giving you more clarity and confidence in its responses.\u201d<\/p>\n Reasoning models like o3-mini thoroughly fact-check themselves before giving out results, which\u00a0helps them avoid some of the\u00a0pitfalls<\/a>\u00a0that normally trip up models. The trade-off is that reasoning models take a little longer to arrive at solutions \u2014 typically seconds to minutes longer.<\/p>\n DeepSeek\u2019s R1 model<\/a>, a \u201creasoning\u201d model along the lines of o3-mini, reveals its full thought process, which many AI researchers argue is the preferred approach. In addition to making the model easier to study, the reasoning steps deliver a better user experience in certain situations, helping indicate when the model might be on the right \u2014 or wrong \u2014 track. <\/p>\n OpenAI had opted not to show the full reasoning steps for o3-mini and its predecessors, o1 and o1-mini, in part due to competitive reasons. Instead, users only saw summaries of the reasoning steps \u2014 summaries that were at times erroneous<\/a>.<\/p>\n When we briefed people on \ud83c\udf53 before o1-preview\u2019s release, seeing the CoT live was usually the \u201caha\u201d moment for them that made it clear this was going to be a big deal. These aren\u2019t the raw CoTs but it\u2019s a big step closer and I\u2019m glad we can share that experience with the world. https:\/\/t.co\/72ZpPrHmfK<\/a><\/p>\n \u2014 Noam Brown (@polynoamial) February 6, 2025<\/a><\/p>\n<\/blockquote>\n OpenAI still isn\u2019t showing o3-mini\u2019s full reasoning steps, but the company said it \u201cfound a balance\u201d: o3-mini can \u201cthink freely\u201d and then organize its \u201cthoughts\u201d into more detailed summaries. <\/p>\n \u201cTo improve clarity and safety, we\u2019ve added an additional post-processing step where the model reviews the raw chain of thought, removing any unsafe content, and then simplifies any complex ideas,\u201d the OpenAI spokesperson continued. \u201cAdditionally, this post-processing step enables non-English users to receive the chain of thought in their native language, creating a more accessible and friendly experience.\u201d<\/p>\n In a Reddit AMA last week<\/a>, Kevin Weil, OpenAI\u2019s chief product officer, hinted that the change was coming.<\/p>\n \u201cWe\u2019re working on showing a bunch more than we show today \u2014 [showing the model thought process] will be very, very soon,\u201d he said. \u201cTBD on all \u2014 showing all chain of thought leads to competitive distillation, but we also know people (at least power users) want it, so we\u2019ll find the right way to balance it.\u201d<\/p>\n TechCrunch has an AI-focused newsletter!\u00a0Sign up here<\/a>\u00a0to get it in your inbox every Wednesday<\/em>.<\/p>\n<\/div>\n\n