OpenAI’s ChatGPT Has Copied DeepSeek In One Key Aspect
Contents
DeepSeek’s growing popularity is causing moves among major rivals in the artificial intelligence space. The Chinese AI platform has some really neat ideas that others could use. It seems that OpenAI has realized this, as ChatGPT o3-mini’s new chain of thought is basically a copy of DeepSeek-R1’s.
As its name suggests, an AI’s chain of thought (CoT) is a breakdown that shows how it arrived at a particular output. It more clearly shows the process where the LLMs “break down” the main task into smaller ones, following a logical sequence. However, ChatGPT had a problem with that: the “raw” CoT data could be too complex. Considering that the company wants its AI platform to be as simple to understand as possible for most users, this is something that probably needed a change.
OpenAI’s revamped “chain of thought” summary is a copy of DeepSeek-R1’s
OpenAI seems to have found inspiration in DeepSeek to solve the problem. The firm has implemented a new approach where CoT is presented in the form of summaries of the reasoning process. This makes things much clearer, although it gets rid of some advanced data along the way. The new way of viewing CoT is present in the ChatGPT o3-mini/o3-mini-high model. It is available to both free users and paid subscribers.
The new way of displaying the chatbot’s CoT is a clear commitment to accessibility. OpenAI claims that the change will “make it easier for people to understand how the model thinks.” The firm says they implemented it because “users have told us that understanding how the model reasons through a response not only supports more informed decision-making but also helps build trust in its answers.”
That said, the new approach has not been without its critics. Users accustomed to getting raw CoT data are complaining that they are now only getting a simplified summary. Meanwhile, others are grateful that the thought process that led to the AI’s output is now much easier to understand. It might encourage more people to use the “Reasoning” option that previously seemed intended for more experienced folks.
New approach also makes CoT easier for non-English users to understand
“While the model’s raw CoT remains hidden as it’s hard to understand, we’ve found a balance: the model can think freely, and then it organizes those thoughts so that they are easy to read. To improve clarity and safety, we’ve 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,” stated OpenAI regarding the change. Another advantage of the new approach is that it “enables non-English users to receive the CoT in their native language, creating a more accessible and friendly experience.”
DeepSeep’s AI models have been surrounded by controversy and concerns about potential risks since its launch. However, it seems that it also brought some good ideas that will help, in one way or another, to make artificial intelligence more accessible.
