ChatGPT and Memory: What a Lawyer Needs to Know
Context and memory influence ChatGPT's responses. Context is limited to the current session, while memory retains details across conversations. For lawyers, temporary chats ensure confidentiality.
When you interact with ChatGPT, two fundamental concepts influence the responses you receive: context and memory.
Context
Every time a new chat is opened, a new context is generated. The context is built from every message sent or received in the chat and is enriched with details that are considered only until the end of the session. The context has a size limit that depends on the type of subscription used.
Memory
If memory is enabled, ChatGPT can keep track of details discussed in previous sessions and reuse them.
Memory refers to the system’s ability to remember information across different conversations. It is a bridge between different chats; it is a bridge between different created contexts.
A practical example
Imagine opening a new chat and saying to ChatGPT: "I’m learning to play the guitar."
Without memory enabled, if you return tomorrow and say, "I bought new strings," ChatGPT will not understand the reference and will ask for clarification.
With memory enabled, ChatGPT will remember that you were learning to play the guitar and will immediately understand the context.
Therefore, memory has a significant impact on the user experience.
For a lawyer
For a lawyer, understanding the distinction between context and memory is crucial to ensuring confidentiality and information security.
If ChatGPT does not have memory enabled, each conversation starts from scratch, reducing the risk of misuse of sensitive data.
If ChatGPT has memory enabled, it is essential to assess which information to share to prevent responses from being influenced by irrelevant details.
Temporary chats
Temporary chats do not affect ChatGPT's memory. When you start a temporary chat, the information shared in that session is only used to generate responses in the moment but is not stored for future conversations.
Even if you are using ChatGPT with memory enabled, interactions in temporary chats are part of the context but will not be recorded in memory.
For a lawyer, using temporary chats is an effective way to prevent AI errors or hallucinations and ensure that the provided information is always contextual, not influenced by past conversations.
At present, there is no default setting to always start a chat in temporary mode on ChatGPT. Each time you open a new session, you must manually select the option to start a temporary chat.
