ChatGPT doesn’t have up-to-date data so be careful how you use it

Stick to using ChatGPT to help you write your emails or make your workouts because its data isn’t up to date.

ChatGPT’s app on iOS is currently trending on TikTok because one of its voices sounds very, very much like Scarlett Johansson. I decided to be my best Joaquin Phoenix self and try it out these past few days, and it’s equal parts impressive and scary. Last night, I asked Scarlett—sorry, I meant the AI—about Matthew Perry. It promptly gave me a brief rundown about the actor’s life, but something interesting caught my ear.

It referred to him in the present tense.

I decided to ask it when Matthew Perry passed away. It replied that as of its last “knowledge update”, he was still alive, because it only has information available up to January 2022.

[Hero image: Jonathan Kemper/Unsplash]

The data of ChatGPT is limited up to a certain time so be careful how you use it

ChatGPT has become a useful tool for many things, but asking it to give you recent information may not be the best. To be fair, it’s not like OpenAI is hiding its limitations, which it clearly lays out on its website. Also, it’s not like ChatGPT was tailor-made to provide you with information, unlike Siri, Google, or Alexa. Those assistants actually scour the internet for information in real time. ChatGPT, meanwhile, isn’t connected to the internet at all. 

The app that can be downloaded on phones is based on the ChatGPT 3.5 architecture, which was only trained with data up until January 2022. ChatGPT 4 data, which you need a subscription to use, is a little more recent and knows info from up until April 2023. Considering it’s already 2024, it’s still quite a big knowledge gap. For example, ChatGPT 4 won’t know the box office earnings of Barbie. ChatGPT 3.5 doesn’t even know the movie exists.

“ChatGPT is not connected to the internet, and it can occasionally produce incorrect answers,” OpenAI clearly states in the “What is ChatGPT?” section of its website. It’s something everyone should keep in mind as they use it. The danger, really, isn’t with OpenAI or ChatGPT, because the company is transparent about the AI’s limitations. The danger is in people who start thinking that what the AI is saying is always 100% factual.

For now, good old Google is still your friend for facts (though you still need to double-check, of course). As for ChatGPT, it’s best to just use it for fun as we try and prove that it’s really Scarlett Johansson all along.