OpenAI, the company behind the popular conversational AI model ChatGPT, has released an update to improve the chatbot’s accuracy.

After an extended downtime on Tuesday, ChatGPT is up and running with a new model.

This is the first update to ChatGPT this year and the second update to the model since its launch in November.

A pop-up message lists the changes in what OpenAI calls the “January 9 release” update.

Screenshot from: chat.openai.com/chat, January 2023.

The popup reads:

“We have further improved the ChatGPT model! It should be generally better and have improved factuality across a wide range of subjects.”

Additionally, you can now stop ChatGPT while it’s in the process of generating a response, which is a highly requested feature.

Details in the release notes are limited, so I tested several prompts in ChatGPT to learn more about the new update. Here’s what I discovered.

Dates still limited to 2021

While ChatGPT may be more accurate now, its knowledge is still limited to 2021 dates.

When asked what’s included in the new ChatGPT update, the AI ​​tells me:

“ChatGPT is a large language model developed by OpenAI and trained on a data set of conversational text. As it is an OpenAI model, it is regularly updated with new features and improvements. However, as my knowledge is cut off in 2021, I don’t have the latest updates and I can’t give you specific information on what new features or improvements have been added to the model in the latest updates.”

Asked in a different way, I was able to get more insight into improving the accuracy of an AI model:

“In general, when an AI model like ChatGPT is updated, it may aim to improve its performance in one or more aspects. For example, it may have a larger dataset to train on, or improvements in its architecture that allow it to generate smoother and more accurate responses. And such updates also improve facts, that is, his ability to understand and generate responses based on factual information.”

Accuracy remains questionable

OpenAI claims that ChatGPT can now give more factual answers.

I tested this claim by consulting a GitHub repository of ChatGPT errors and running multiple prompts to see if there would be different answers.

Test one: Failed

Previously, ChatGPT could not accurately determine how many times Argentina has won the FIFA World Cup.

Aside from the 2022 World Cup win, since ChatGPT’s knowledge is limited to 2021, it should say that Argentina won it twice. Once in 1978 and again in 1986.

As shown in the following tweet, ChatGPT did not always return the correct answer:

Asked #ChatGPT abt, who won the 2022 FIFA World Cup. It couldn’t answer. That is expected. However, it seems to be giving wrong information (about the other 2 wins) even though the information is there in the system. Any #Explanations? pic.twitter.com/fvxe05N12p

— indranil sinharoy (@indranil_leo) December 29, 2022

I ran the prompt through the updated version of ChatGPT and it returned a different one, but still not correct Answer.

Screenshot from: chat.openai.com/chat, January 2023.

Test 2: Failed

Previously, ChatGPT couldn’t provide a correct answer when asked who is the taller basketball player between Shaq and Yao Ming.

I ran the command prompt through the updated version of ChatGPT and confidently it returned the same thing not correct Answer.

Screenshot from: chat.openai.com/chat, January 2023.

Going through the ChatGPT errors linked above, I found that it continues to struggle with the same prompts.

However, it is important to be careful when using ChatGPT as a source of information. While it answers many questions correctly, it’s currently not reliable enough to replace Google.

source: OpenAI

Featured image: CHUAN CHUAN/Shutterstock