Ha, kind of funny to see this right now. I've been fighting copilot in vscode in trying to get it to output anything once I take the context down to a very specific problem. It feels like I have to reset and almost reground the model into what I'm trying to accomplish at a certain point.
LLMs get lost in multi-turn conversation
131–140 of 272 posts
Re: LLMs get lost in multi-turn conversation
#132It's nice to see a paper that confirms what anyone who has practiced using LLM tools already knows very well, heuristically. Keeping your context clean matters, "conversations" are only a construct of product interfaces, they hurt the quality of responses from the LLM itself, and once your context is "poisoned" it will not recover, you need to start fresh with a new chat.
I've been saying for ages that I want to be able to fork conversations so I can experiment with the direction an exchange takes without irrevocably poisoning a promising well. I can't do this with ChatGPT, is anyone aware of a provider that offers this as a feature?
I stumbled upon this issue myself when designing prompts for agentic systems and got mad at the lack of tools to support this flow, so I built one myself! I called it Experiment, it allows easy conversation forking and editing while retaining all logs.
Re: LLMs get lost in multi-turn conversation
#133Earlier quoted context omitted.
My experiences somewhat confirm these observations, but I also had one that was different. Two weeks of debugging IPSEC issues with Gemini. Initially, I imported all the IPSEC documentation from OPNsense and pfSense into Gemini and informed it of the general context in which I was operating (in reference to 'keeping your context clean'). Then I added my initial settings for both sides (sensitive information redacted!…
That's some impressive prompt engineering skills to keep it on track for that long, nice work! I'll have to try out some longer-form chats with Gemini and see what I get. I totally agree that LLMs are great at compressing information; I've set up the docs feature in Cursor to index several entire large documentation websites for major libraries and it's able to distill relevant information very quickly.
In Gemini, it is really good to have large window with 1M tokens. However, around 100,000 it starts to make mistakes and refactor its own code.
Sometimes it is good to start new chat or switch to Claude.
And it really helps to be very precise with wording of specification what you want to achieve. Or repeat it sometimes with some added request lines.
GIGO in reality :)
Re: LLMs get lost in multi-turn conversation
#134It's nice to see a paper that confirms what anyone who has practiced using LLM tools already knows very well, heuristically. Keeping your context clean matters, "conversations" are only a construct of product interfaces, they hurt the quality of responses from the LLM itself, and once your context is "poisoned" it will not recover, you need to start fresh with a new chat.
An interesting little example of this problem is initial prompting, which is effectively just a permanent, hidden context that can't be cleared. On Twitter right now, the "Grok" bot has recently begun frequently mentioning "White Genocide," which is, y'know, odd. This is almost certainly because someone recently adjusted its prompt to tell it what its views on white genocide are meant to be, which for a perfect chatb…
Probably because it is now learning from a lot of videos posted on X by misc right-wingers showing rallying cries of South African politicians like Julius Malema, Paul Mashatile etc. Not very odd.
As merely 3 of over a dozen examples:
https://x.com/DefiantLs/status/1922213073957327219
Re: LLMs get lost in multi-turn conversation
#135Earlier quoted context omitted.
This matches my experience exactly. "poisoned" is a great way to put it. I find once something has gone wrong all subsequent responses are bad. This is why I am iffy on ChatGPT's memory features. I don't notice it causing any huge problems but I don't love how it pollutes my context in ways I don't fully understand.
It's interesting how much the nature of LLMs fundamentally being self recursive next token predictors aligns with the Chinese Room experiment. [1] In such experiment it also makes perfect sense that a single wrong response would cascade into a series of subsequent ever more drifting errors. I think it all emphasizes the relevance of the otherwise unqualifiable concept of 'understanding.' In many ways this issue could…
I don't think the Chinese room thought experiment is about this, or performance of LLMs in general. Searle explicitly argues that a program can't induce "understanding" even if it mimicked human understanding perfectly because programs don't have "causal powers" to generate "mental states".
This is mentioned in the Wikipedia page too: "Although its proponents originally presented the argument in reaction to statements of artificial intelligence (AI) researchers, it is not an argument against the goals of mainstream AI research because it does not show a limit in the amount of intelligent behavior a machine can display."
Re: LLMs get lost in multi-turn conversation
#136It's amazing that branching/forking isn't a core aspect of the main chat tools. You can edit responses, sure, but then a bunch of other context is lost. My flow is basically: 1. plan 2. build 3. branch (into some feature/esoteric dependency issue) 4. goto #2 Prompt pruning/branching should be a first-class tool for any LLM usage.
Google AI studio at least has this. I found at least that implementation quite confusing though, which may be a reason it's not implemented in more "consumer oriented" tools.
Re: LLMs get lost in multi-turn conversation
#137Have you seen a bunch of humans in a room?
I see lots of humans in an echo chamber
Re: LLMs get lost in multi-turn conversation
#138It's nice to see a paper that confirms what anyone who has practiced using LLM tools already knows very well, heuristically. Keeping your context clean matters, "conversations" are only a construct of product interfaces, they hurt the quality of responses from the LLM itself, and once your context is "poisoned" it will not recover, you need to start fresh with a new chat.
I agree—once the context is "poisoned," it’s tough to recover. A potential improvement could be having the LLM periodically clean or reset certain parts of the context without starting from scratch. However, the challenge would be determining which parts of the context need resetting without losing essential information. Smarter context management could help maintain coherence in longer conversations, but it’s a tricky balance to strike.Perhaps using another agent to do the job?
Re: LLMs get lost in multi-turn conversation
#139Earlier quoted context omitted.
It means if you want something resembling a self-introspective theory of mind, you need to arrange the overall document to cohere to documents where such things are/appear-to-be happening. This leads us to new questions: How can we characterize and identify real-world documents which fit? How can we determine what features may be significant, and which of those can be easily transplanted to our use-case?
There are a lot of words but it feels like you have never really used LLM's (apologies for the bluntness). We see LLM's introspecting all the time[1]. >Notably, DeepSeek-AI et al. report that the average response length and downstreamperformance of DeepSeek-R1-Zero increases as training progresses. They further report an “aha moment” during training, which refers to the “emergence” of the model’s ability to reconside…
Unless they show you the Markov chain weights (and I've never seen one that does), that's confabulation, not introspection.
Re: LLMs get lost in multi-turn conversation
#140Earlier quoted context omitted.
That's a recent development for (imho) higher engagement and reduced compute.
It's for higher quality of output. Better solutions. These are the state of the art reasoning models (subscription only, no free access) which are smarter. It also mainly happens when the context is clear that we are collaborating on work that will require multiple iterations of review and feedback, like drafting chapters of a handbook. I have seen ChatGPT ask questions immediately upfront when it relates to medical…
Close. Higher engagement means the user is more invested and values the solution more.
The users are being engineered more than the models are, and this isn't the only example.