The concept of memory and more when talking about a service like ChatGPTit can cause surprise and cause feardepending on its development. And therefore, of course, it is essential to describe in detail what it consists of, and of course what controls the user has over this feature that OpenAI has just announced and which, at least a priori, seems quite interesting, and what is even more important, offers an adequate level control over its operation.

But let’s start from the beginning What does OpenAI mean when it talks about adding memory to ChatGPT? From what we can read in your publication, we’re talking about a feature that will allow the chatbot to remember specific aspects of conversations we’ve had with it before and take them into account in all conversations from that point on. And yes, the key point here is that we’re not talking about memory within the same conversation, but rather pieces of information that the service will store and take into account in all our interactions with it.

For those who use the service more or less regularly, may remind you of the Customize ChatGPT feature, which in the long term allows us to establish some parameters that will be taken into account in the interviews. With this new memory, the possibility seems to go much further, as ChatGPT will allow many very specific aspects of our context to be remembered and always taken into account, even if we do not recall it.

Although we can access some aspects of memory management from the ChatGPT settings, we will do the next part of it directly in the interviews. To this end, we may tell you that we want you to remember something we have just told you, we may also ask you to list any memories you have stored through this system, and if we wish, we will be able to tell you, that we want to erase the memory in particular. This is very important because it makes it clear to us that the service will remember what we want it to remember, not what it thinks it should remember.

From the service configuration, we can activate and deactivate the memory function, and also perform its complete removal, if we have added too many memories and want to clean up. At the moment, OpenAI has deployed this feature to a small number of service accounts, and it seems that its scope will be complete, that is, it will reach both paid and free accounts. Of course, once the feature is available for an account, it will be enabled by default, so if you don’t want to use it, you’ll need to disable it in the settings.

I personally find this concept very interesting., although I think it would be improved if OpenAI offered an alternative method to conversations to view the history of ChatGPT memories, as well as the ability to delete the ones we don’t want to keep anymore. And I’m also wondering if something like this will be necessary to avoid bigger problems for the service in Europe.