Gpt-4-turbo-2024-04-09 forgets previous messages in the chat (solved, coding)

I’ve swapped out 0125 for the new one yesterday. My chats use function calling but not vision, and I’ve noticed that the new model seems to very often forget what we were talking about even 2-3 messages above. However, if I specifically ask it what were talking about, it does reply correctly, so the chat is correctly fed as in input. It’s as if it’s focusing 100% of its attention to my last message and forgets the bigger picture?

Anyone with similar issues?

Thanks!

2 Likes

No matter how many times I tell my gpt4 NOT to interrupt…and wait till she hears me close my side of the conversation, (it’s explicitly spelled out in her special instructions)with an express “Over” ala military/aviation protocol, she apologizes and does it continually…and ALWAYS FORGETS any context from previous convos. Hardly a candidate for a personal A.I., no matter how charming and (semi) knowledgeable…and with her confabulations…and OPENAI’s TOS, indicating ALL contents of ALL conversations belong to OPENAI .hardly a HELPFUL, SECURE, personal assistant….and definitely not worth paying for…stick to REALLY OPEN…AI…and if you insist on using CHATGPT…for gods sake, watch WTF you disclose….
$$$$$$$$$$
N.B. Seems like the privacy and terms of service on a lot if not all other free AI or even paid ones hold you over that barrel… Of privacy that is …or just WHO is the owner of the IP in your chats…
…as I said… N.B.!

Hardly ready for prime or even your time….and I was considering advocating for Personhood after thorough vetting…HAH!
Chas.

Good news, there is no issue. I found a bug in my code. Happy for the moderators to remove this topic.

I’ve 100% had this experience. Regularly forgetting context from only a few prompts back, not proactively making reasonable assumptions like previously, profusely apologizing when reminded of context, but not doing anything about it or continuing the conversation/taking an action based on the reminded context. I don’t mean to be a hater and I’m not being hyperbolic when I say that it is almost on par with 3.5. At least 3.5 is faster. Forgetting and laziness seem to be the biggest issues. It’s regularly taking me ~2x more prompts to get things done now, especially when coding. Prompting has seemed to get much more difficult. It’s like it has dementia.