[flagged]
The meaningful distinction between lying and wrong, is that a liar doubles down on their position, making up more things and tailoring their statements to what their audience wants to hear. The article presents good arguments that LLMs actually do this. I've observed it myself interacting with ChatGPT. Sure, this seems weird but it's not. Read the article for why.
They are detached from correctness because they're answering a question that is loosely correlated with correctness. It's not like they're attempting to be correct, or care about correctness
All of this seemingly avoids the real questions either side of this poses. Is this technology actually useful to humanity? If not, how much expense is required to make it useful? Finally, would that be better directly invested into people?
I find solving problems and facing challenges to be the most satisfying human activity. I honestly can't comprehend the miles of annoyed apologia written in favor of 'talking' to ChatGPT.
So do intemperate comments ridiculing other positions, it seems.
See HN faq, I think.
That's lying. If you don't know the answer, the correct answer is "I don't know", or to ask clarifying questions. To pretend that you know when you don't, is to lie. Even if your intent is to deceive because you think it's friendly, your intent is to deceive.
> That's equivalent to an LLM, it is mid-thought, stuck at the most recent batch of training samples it was given when you come along and you interrupt it with some new thought.
For someone who thinks hot-takes anthropomorphizing ChatGPT and LLMs shouldn't be given any serious attention, you sure are anthropomorphizing LLMs...
The thing is, LLMs aren't doing that at all. They're just pattern matching prompt data to training data and then mixing together a response from the "continuations" of the matched training data. Nothing is being "interrupted". They're sitting around waiting for prompts.
LLMs can't lie because they can't intend to deceive. They can't intend at all. They have no intent, no will. These are anthropomorphic qualities you're putting on them, which they simply don't have.
But, I think the original article IS NOT arguing that LLMs are capable of lying. It's arguing that the average human doesn't understand what's going on with LLMs, and isn't going to go through the effort to gain that understanding. That's not a judgment: the average person has better things to do with their time than learn all the necessary prerequisite knowledge to understand LLMs. It's a hard problem, because we're trying to explain something really complex to people who don't even understand the building blocks the complexity is made out of. The question the original article is answering is: what's an effective metaphor to explain LLMs to laypeople?
The approach being suggested by the OP to solving this problem is to use an anthropomorphic analogy which gives people the understanding of whether to trust information gained from LLMs. You tell them that LLMs are lying. Yes, we understand that they aren't literally lying--as I said before, LLMs don't have intent so they can't intend to deceive. It's a metaphor.
Telling people that LLMs are wrong sometimes gives the wrong impression. Most people are wrong sometimes, but people are usually trying to be honest and will usually try to not mislead you when it's important, and even when people might be wrong it's worthwhile to take the risk of trusting people because the alternative is living in fear and loneliness. So telling people LLMs are wrong sometimes, gives people the idea that LLMs make mistakes, but they're trying their best so you can trust them. Which is not accurate: LLMs are as biased and confused as the humans that created their training data, but on top of that, have a lot less variety of data, and sometimes randomly combine things in the wrong ways. LLMs can't generally be trusted: they must be generally distrusted, especially on anything where the facts matter.
Telling people that LLMs lie produces the correct general impression. People don't trust a known liar. They might listen to a liar, but they'll take what the liar says with a grain of skepticism. And that's exactly how we should be treating what LLM's say, given their current capabilities.
My thoughts were reinforced that in common American English we refer to both with the same name frequently.
However they are different, and I eventually was corrected and I was able to verify that correction.
The question is, when I legitimately believe my knowledge was correct, was I lying?
There are grey areas here, like levels of confidence in our knowledge. A rational person believes very few things with 100% certainty, but it's not reasonable to be paralyzed by even the slightest amount of doubt and never say anything. I think a reasonable approach is to state things with some indication of your level of confidence, like:
1. Obviously sweet potatoes are yams. 2. Sweet potatoes are yams. 3. I am pretty sure sweet potatoes are yams. 4. I think sweet potatoes are yams. 5. I think sweet potatoes are yams, but I'm not sure. 6. I think sweet potatoes are yams, but I don't really know why I think that. 7. Sweet potatoes might be yams.
You can use that definition, but it is not how the word "lie" is commonly used.
Mirriam-Webster defines lie as "to make an untrue statement with the intent to deceive"[0]
Cambridge dictionary defines lie as "to say or write something that is not true in order to deceive someone"[1]
Colllins dictionary defines lie as "A lie is something that someone says or writes which they know is untrue." and "If someone is lying, they are saying something which they know is not true."[2]
It's sad that people are so willing to conflate lying with being wrong to push a narrative.
0. https://www.merriam-webster.com/dictionary/lie
1. https://dictionary.cambridge.org/us/dictionary/english/lie
2. https://www.collinsdictionary.com/us/dictionary/english/lie
The problem is that any word that ascribes agency to the LLM will technically be incorrect. But that removes most possible descriptions of its tone and style which are a relevant part of its response.
An analogy would be if ChatGPT started insulting me and calling my question stupid. Would it be wrong to call its response "rude" or "mean" just because it is statistically regurgitating text that matches some input parameters? This seems unreasonable if our goal is to capture the gist of its response.
This is why people judge it to be "lying" rather than being merely incorrect: it is responding with a certain conversational tone in a certain context that gives its answer a style of arrogance, deceitfulness, and narcissism (because I guess that's what internet comment boards are filled with). "Lying" is a description of the totality of its response--including tone and style--not just the truth value of the answer.
If we are to be really pedantic, the LLM isn't even correct or incorrect ;it is just completing strings of tokens. Humans are imputing their own judgment about what those tokens mean--same as with tone and style. Imputing tone isn't so different from imputing truth value.
It’s important because of motive. A lie is told with intent to conceal a known truth. It’s not just LLM agency that’s in question, it’s human malice.
Co-opting the term “lie” is a rhetorical tool used to shift the conversation from “this person/LLM is saying things that aren’t true” to “this person/LLM is acting maliciously and needs to be punished/sanctioned”
All I want is an honest conversation. It’s more than a little ironic that many humans are themselves being less than honest (and sometimes just bullshitting!) about the context when LLMs produce false statements.
> It’s important because of motive. A lie is told with intent to conceal a known truth. It’s not just LLM agency that’s in question, it’s human malice.
That is not a distinction that matters to me with humans or AI. "Bullshitting" is just a specific form of lying. "Lying" doesn't imply malice to me. But a lack of malice is fairly irrelevant: mostly when people say they didn't intend any harm by lying, it's true, but it's just a way to minimize the fact that they lied. Intent is not as important as you seem to think it is: intent has little to do with how much harm is caused by a lie.
Intent is particularly irrelevant with AI, because AI doesn't have intent. Contrary to your statement, LLM agency isn't in question: we know AI does not have agency.
> Co-opting the term “lie” is a rhetorical tool used to shift the conversation from “this person/LLM is saying things that aren’t true” to “this person/LLM is acting maliciously and needs to be punished/sanctioned”
That's a pretty big leap. I'm not saying that AI should be punished or sanctioned. In fact, I'm not even sure what it would mean to "punish" something that can't experience suffering. What I want isn't to punish/sanction AI, what I want is for people to not trust AI.
I agree and think there probably is a missing term in our language for the phenomenon we are observing, but it feels like splitting hairs to say the LLM is merely bulshitting and not lying. For the average person dealing with ChatGPT this distinction won't matter, and saying ChatGPT sometimes "lies" more clearly communicates the possible negative downside of its answer than saying it sometimes "bullshits" (especially for low literacy or non-native speakers).
I mean, isn't bullshitting still a kind of lie? It's an implicit lie that one is qualified and intends to speak the truth. Certainly it is a kind of deception about one's qualifications, even if that is self-deception. It seems like we are just arguing about shades of gray when it's unclear why that matters.
If I don't believe I know something, and I pretend to know it, the untrue statement is implicit: I'm stating that I have information that I don't have.
For example, I have heard that there's a town called "Springfield" in every state, but I've never done any research to verify that. If I confidently say, "There's a town called 'Springfield' in every state," that's a lie, even if there IS a town called 'Springfield' in every state. The deception isn't the geographical information, it's the implicit statement that I know this geographical information when I don't.
> It's sad that people are so willing to conflate lying with being wrong to push a narrative.
I'm not conflating those things. I think you'll find that what I'm describing fits the definitions you've given, and I think I've made it very clear that I do understand that intent is a necessary component of lying.
In any case, nitpicking terminology misses the point being made.
I agree w your take.
https://www.google.com/search?client=firefox-b-d&q=%22anthro...
5000 hits
'Broken' or 'wrong' would be fine if that were simply just making an error.
But if there's something in the system, even the corpus, that permeates a falsehood then I think 'lying' is the right term.
ChatGPT has a morality, and it's a bit specific to the Anglosphere. It talks like the perfect LinkedIn corporate bot. This means that it will effectively misrepresent ideas on a lot of sensitive subjects from the purview of many people in the world. That's a form of 'structured misrepresentation' or maybe just bias, but it can feel like lying.
'Hallucinating' is a cool sounding but inappropriate anthropomorphization.