Like a lot of people I have been using AI platforms to help manage large amounts of data and information as it makes things a lot easier. Hallucination for these AI platforms is a common occurrence and so you need to ensure what they are telling you is accurate. What was interesting though was I observed how with one platform it seemed like I had to keep reminding this AI about certain points when clearly the data about the subject was entered. You would expect a human to be forgetful but not a machine.
I then got the AI to explain to me how it has a memory limit where it can only recollect maybe twenty of the last messages in a long conversation and that when you are talking later on it’s not actually remembering the exact data it had before in answer questions but rather it is guessing based on the current conversation. It’s similar to when you tell a buddy about an event they don’t remember as they just nod their head and then insert whatever you just said as the fact. I even question it on how I can scroll up and see the old data but it reaffirmed the AI can’t actually remember something that far back in the conversation.
However, it did suggests platforms that could retain a large amount of data. This is very important to know if you are using an AI to do say a case study of sort where you dump it all this information only to find out it can’t remember everything as you are chatting away with it. This kind of explains a lot too with things like hallucination. Be careful how you use it.
