ChatGPT Enhances Math Capabilities with Update

January 31, 2023

OpenAI has released an update to its popular language model, ChatGPT, to enhance its accuracy and improve its ability to handle math equations. According to their January 30th release notes, the new update is designed to give ChatGPT improved factuality and mathematical capabilities, meaning that it should be better able to provide accurate information, calculations, and answers for students, researchers, and professionals who need quick and reliable data. While the new version of ChatGPT is still far from perfect in terms of equation handling, it has shown significant improvements when it comes to providing factual responses.

The January 30 update to ChatGPT has brought some positive improvements in terms of accuracy, as it is no longer possible to trick it into giving an incorrect response. This was previously demonstrated by a meme showing how saying "my wife disagrees" could get ChatGPT to give the wrong answer (see this tweet for an example). Following the update, this loophole has been closed and ChatGPT continues to return the correct response even if it is challenged. A test conducted following the update came back with a positive result (screenshot here). However, feedback on the January 30 update appears to be largely negative in comparison to these improvements.

ChatGPT, an AI-powered chatbot, is known to struggle with tasks that involve mathematical equations expressed in words rather than numbers and symbols. An example of this is asking the chatbot which basketball player, Shaquille O’Neal or Yao Ming, is taller. Despite returning the correct heights of the two men, ChatGPT was unable to answer the question correctly until it was pointed out. After its Jan 30 update promised “improved factuality and mathematical capabilities” however, it still failed in a previous failure mode where it provided incorrect answers for related math questions on Twitter. According to Deedy (@debarghya_das), it gave 44% as an answer for a quantity that should have been 46%, and 1555.8.. instead of the correct 1551.9.. (

On a positive note however, ChatGPT performs exceptionally well when fed questions from standardized tests such as the SAT math section - as one individual claims to have successfully passed by with assistance from ChatGPT.

OpenAI recently released an upgraded version of their ChatGPT model, with improved mathematical capabilities. Charis Zhang, a member of the OpenAI team, tested the newest version on a 2020 SAT and found that it was able to solve two problems correctly in under five seconds. The enhanced math capabilities of ChatGPT may be due to its exposure to similar language seen in standardized tests, rather than user-inputted questions which it may not have encountered before.

The release of this update marks the third major update since the introduction of ChatGPT and reflects OpenAI's commitment to staying ahead in the AI industry. Feedback on this update is mixed; while ChatGPT has made important advances with this upgrade, further testing is needed before fully relying on its responses.

For more information about OpenAI's upgraded ChatGPT model and Charis Zhang's experiment with it, see here. OpenAI, a research laboratory dedicated to discovering and developing cutting-edge artificial intelligence technologies, has recently released an update for their open-source natural language processing model ChatGPT. This update follows the previously established schedule for OpenAI, with further improvements expected in the near future. ChatGPT is an AI-driven natural language generator that leverages a large pre-trained model to generate conversational responses based on user input. It is designed to enable natural conversations between users and AI agents. This update brings improvements such as better sentence completion, increased accuracy in sentence structure, and more natural conversation flows. With these improvements, developers can now create more realistic conversation experiences with their chatbots and other AI applications. OpenAI continues to be at the forefront of developing advanced artificial intelligence technologies for various uses. Be sure to stay tuned for further updates on ChatGPT and other OpenAI projects!

Exploring Yandex Search Ranking Factors with Leaked Insights
The search marketing community is trying to make sense of the leaked Yandex repository containing files listing what looks like search ranking factors. Ryan Jones (@RyanJones) believes that this leak...
Read More
OpenAI Introduces Tool For Identifying AI-Generated Content
OpenAI, the AI research firm behind ChatGPT, has released a new tool – the AI Text Classifier – to help distinguish between AI-generated and human-generated text. This tool is intended to help mitigat...
Read More