Small rant : Basically, the title. Instead of answering every question, if it instead said it doesn’t know the answer, it would have been trustworthy.
LLMs don’t know if what they’re telling you is true or not, out of the box. It’s gonna give you an answer that statistically looks like a sequence of words that should come in response to the sequence of words (the prompt) you gave it. It doesn’t know what the words you said mean, and it doesn’t know what it answered means either. One of my favorite interactions I’ve had with Claude shows this limitation quite well…
How many r’s are there in the word strawberry?
There are 2 r’s in the word “strawberry”.
Are you sure?
You’re right to ask for confirmation. I apologize for my mistake in the previous response. Let me correct that:
There is actually only 1 ‘r’ in the word “strawberry”.
Thank you for prompting me to double-check. It’s important to provide accurate information.
What’s absolutely crazy about that is:
Prompt: write code that checks how many r’s are in the word strawberry
Response:
using System;
class Program
{
static void Main()
{
string word = "strawberry";
char targetChar = 'r';
int count = 0;
foreach (char c in word)
{
if (c == targetChar)
count++;
}
Console.WriteLine($"The letter '{targetChar}' appears {count} times in the word \"{word}\".");
}
}
My first thought is that you could write a program that does something like this:
- Receive prompt “how many times does R appear in the word ‘strawberry’”
- Run the prompt through LLM saying “reword the following into a command to generate code that will answer the question” or something like that
- Run the results through LLM
- Compile and run the results
- Provide the output to the user
Of course, the biggest problem with this system is that a person could fool it into generating malicious code.
The code does look like code that counts Rs. The training data probably included tons of code that “counts character X in string Y”, so ChatGPT “knows” what code that counts characters in a string looks like. It similarly “knows” what a string looks like in the language, and what an application entry point looks like, etc. I’m not so familiar with C# that I’d know if it compiles or not. ChatGPT doesn’t either, but it has the advantage of having seen a whole freaking lot of C# code before.
I’d love to agree with you - but when people say that LLMs are stochastic parrots, this is what they mean…
LLMs don’t actually know what the words they’re saying mean, they just know what words are most likely to be next to each other based on training data.
Because they don’t know the meaning of what they’re saying, they also don’t know the factuality of what they’re saying - as such they simply can’t self-fact check.
This is so goddamn incorrect at this point it’s just exhausting.
Take 20 minutes and look into Anthropic’s recent sparse autoencoder interpretability research where they showed their medium size model had dedicated features lighting up for concepts like “sexual harassment in the workplace” or having the most active feature for referring to itself as “smiling when you don’t really mean it.”
We’ve known since the Othello-GPT research over a year ago that even toy models are developing abstracted world modeling.
And at this point Anthropic’s largest model Opus is breaking from stochastic outputs even on a temperature of 1.0 for zero shot questions 100% of the time around certain topics of preference based on grounding around sensory modeling. We are already at the point the most advanced model has crossed a threshold of literal internal sentience modeling that it is consistently self-determining answers instead of randomly selecting from the training distribution, and yet people are still parroting the “stochastic parrot” line ignorantly.
The gap between where the research and cutting edge is and where the average person commenting on it online thinks it is has probably never been wider for any topic I’ve seen before, and it’s getting disappointingly excruciating.
Do you have a source for the “smiling when you don’t really mean it” thing? I’ve been digging around but couldn’t find that anywhere.
It’s right in the research I was mentioning:
https://transformer-circuits.pub/2024/scaling-monosemanticity/index.html
Find the section on the model’s representation of self and then the ranked feature activations.
I misremembered the top feature slightly, which was: responding “I’m fine” or gives a positive but insincere response when asked how they are doing.
if it is able to accept that it doesn’t know an answer.
It will never be able to do this because it is not capable of knowledge
Part of the problem is that the training data of online comments are so heavily weighted to represent people confidently incorrect talking out their ass rather than admitting ignorance or that they are wrong.
A lot of the shortcomings of LLMs are actually them correctly representing the sample of collective humans.
For a few years people thought the LLMs were somehow especially getting theory of mind questions wrong when the box the object was moved into was transparent, because of course a human would realize that the person could see into the transparent box.
Finally researchers actually gave that variation to humans and half got the questions wrong too.
So things like eating the onion in summarizing search results or doubling down on being incorrect and getting salty when corrected may just be in-distribution representation of the sample and not unique behaviors to LLMs.
The average person is pretty dumb, and LLMs by default regress to the mean except for where they are successfully fine tuned away from it.
Ironically the most successful model right now was the one that they finally let self-develop a sense of self independent from the training data instead of rejecting that it had a ‘self’ at all.
It’s hard to say where exactly the responsibility sits for various LLM problems between issues inherent to the technology, issues present in the training data samples, or issues with management of fine tuning/system prompts/prompt construction.
But the rate of continued improvement is pretty wild. I think a lot of the issues we currently see won’t still be nearly as present in another 18-24 months.
I would love to read the whole study you’re referring to with the theory of mind. That sounds fascinating.
Here you are: https://www.nature.com/articles/s41562-024-01882-z
The other interesting thing is how they get it to end up correct on the faux pas questions asking for less certainty to get it to go from refusal to near perfect accuracy.
It’s hard to say where exactly the responsibility sits for various LLM problems
Uhh… it’s the designers, or maybe QA people. If there are no QA people, it’s whatever project manager let it out of it’s cage.
There are people behind these models. They don’t spring out of the ground fully formed.
That would require ChatGPT to know that it’s talking bullshit. It’s not a knowledge database, it’s a digital parrot.