Video Gamer is reader-supported. When you buy through links on our site, we may earn an affiliate commission. Prices subject to change. Learn more
Chatbots are getting better and better at understanding us, so is ChatGPT learning from users?
AI is improving at a staggering rate. Just recently OpenAI announced various plugins for their chatbot. One of which, the code interpreter, is an incredibly powerful tool for programming. What’s more, the new GPT-4 is available for ChatGPT Plus subscribers and it is far more capable than the already impressive GPT-3.5 model that can be used for free.
What’s clear from this is that ChatGPT is improving exponentially, so it is fair to ask whether it is learning from us.
Originality AI detector
Does ChatGPT learn from users?
The basic answer is yes, but you do have some control over how much. For example, some of the ways ChatGPT learns from you are optional. If ChatGPT gives an incorrect response or misunderstands your question you have the option to ‘thumbs down’ the response on the right-hand side of the screen.
If you do this a text box will then appear. In it, you can provide additional information or tick a box to explain your issue with the response. This feedback will be available to OpenAI and used to improve ChatGPT.
However, OpenAI says that they still may access your conversations with ChatGPT, even if you do not flag a specific interaction. They state in their FAQs that this is both safeguarding and training.
Because of this, they advise you to never tell ChatGPT personal or private information, though they do you that all your information is encrypted and your data is kept anonymous.
You also have the option to remove your data from OpenAI entirely by going through their data deletion process.
However, any change to ChatGPT, maybe from using data input by users, only comes into effect after an update. ChatGPT itself is not changing from conversation to conversation, constantly learning, but improving in bigger steps through these updates.
Frequently asked questions
Is my data safe with ChatGPT?
OpenAI makes clear that you should never type personal or private information into ChatGPT. This is both in case of potential security issues and because OpenAI can access your conversations to train ChatGPT.
Is ChatGPT biased?
Because ChatGPT is trained on data and language created by humans, it can be prone to expressing the same biases we do, some of them problematic. In fact, the creators of ChatGPT have admitted much and are committed to fixing this problem.