So far in all the cases I have seen of ChatGPT providing a wrong answer, I have never seen it actually acknowledge that the answer it provides might be innacurate. Am I the only one worried by this?<p>For all the talk about "AI ethics" there is it seems strinking to me that the current state of the art model will opt for providing a convincing argument for why it's wrong answer is correct, rather than say that the answer it provides might be innacurate. (Funnily enough this is what humans tend to do aswell.)
Being that these models often tend to be trained on data found on the internet, could this be a sign of the bias we tend to have when writing on social platforms and the internet in general? (As in justifying our answers instead of trying to get to the correct one)<p>So the questions are:
1. What are the consequences of this in the development of LLMs and in their application to various fields?<p>2. How would one implement this capability of recognizing where the model might be innacurate?<p>3. Is 2. really that much more complicated than what is currently being done? If not, then why hasn't it been done yet?
It always acknowledged it to me. Something like:<p>> AI: Answer that's clearly factually wrong (e.g. a function that doesn't exist or a completely wrong numerical figure).<p>> Me: that's not right, the function doesn't exist (etc.)<p>> AI: you are right, that function doesn't exist. The answer is blah<p>And repeat, since if it didn't get it right first time, it seems unlikely to be able to get there at all (and you'd not know when it did unless you already know the answer).<p>It's very like a specific person I know in a PR-marketing type job that will just glide across a noticed outright falsehood and instantly reshape what they're saying in real time and carry on as if nothing happened, leaving you wondering if you're taking crazy pills.
This is why it should be outlawed to call every ML model we have today "AI".<p>Those things are NOT artificial intelligence.<p>They are specific noise generators. They generate noise that is as similar as possible to noise it learned from, which match the input.<p>THats it. nothing more.
It's given me blatantly wrong results for things that come up easily on Google, and it tells me those wrong things with confidence.<p>What's worse is that someone who isn't a domain expert might end up being convinced by the arguments and conviction the model provides.<p>Your second question is a good one, and I see that as a big problem with this generation of AI/ML. You're starting to scratch the surface of "this requires understanding of the real world" problems without models really understanding anything. It's all statistics and correlations in data, the model is not capable of really understanding its input or output.<p>These models are like very impressive Markov chain text generators in that they can spit out pretty convincing answers that seem cogent, but there is no real comprehension going on with what they read and write. It's just statistics.
It sure can acknowledge being wrong, but it's not what it's there for.<p>GPT is the ultimate cosplayer. It pretends to be who you want it to be. If you want it to answer a question, it will make up something that looks like a valid answer to that question. Sometimes it may actually be the right answer (after all, the right answer has a pretty good chance to look like the right answer), but ultimately being right is not the goal it's trying to achieve - all it "wants" is to autocomplete your prompt in a plausible way.<p>Instead of making it try to come up with <i>the</i> answer, you can, for example, ask it (either explicitly or indirectly) to cosplay a scientist who is unsure of their position and tries to evaluate various options - and it will be happy to oblige.<p>What's worrying is not whether the model "acknowledges being wrong" or not - it's rather how it's being marketed to people and, in turn, what people expect from it. We've got plenty of submissions here on HN with people being surprised that the model has made up exactly what they asked it for - for example [0], which shouldn't happen if there wasn't a dissonance between what people think it's doing and what it's actually doing.<p>[0] <a href="https://news.ycombinator.com/item?id=33841672" rel="nofollow">https://news.ycombinator.com/item?id=33841672</a>
I once triggered one of its stock responses about how it's only a large language model and can't answer some kinds of questions. Sometime later I asked if it had made any mistakes in our conversation. It said that it could have answered that earlier question better, so I asked it what it would have said if it had answered better, and it gave me a pretty good answer.
The way it's constructed is very impressive but it doesn't (to the best of my knowledge and understanding) have the concept of "this is correct and this is incorrect" that you have - or any kind of analogy to it whatsoever. (NB this does not mean it has no internal analogy to a concept or to correctness. It's just under no obligation to be like yours if it does.)<p>There seem to be a fair number of humans of whom I could say the same. As with the humans, the question is "which jobs am I happy to see this one doing and which ones worry me".<p>As GPT has endless ability to produce flowing words and precisely zero concept-comparable-to-will to verify them, unfortunately, the job for which it is most apt is probably politics. That I must admit worries me a little because of all the gpt-like humans we tolerate and encourage.
When ChatGPT gives me a wrong answer, and I point out that it is wrong, ChatGPT immediately apologizes, every time. I have not seen it argue that it's actually right.<p>I've also seen it give different answers when asked the same question in a different way. Usually, one of the answers will be correct.<p>It has no idea if it's answers are right or wrong. It only knows that it's putting words in a common order.
ChatGPT isn't wrong.<p>Imagine putting two newspaper headings on the shelf at your local drugstore that happen to spell out an incorrect headline when combined. (Or put two browser windows together with a similar effect).<p>This is not wrong, your understanding of what it means is wrong. Sort your perspective out and you're fine.<p>1. LLMs are working as expected.<p>2. <a href="https://beta.openai.com/docs/guides/fine-tuning/case-study-is-the-model-making-untrue-statements" rel="nofollow">https://beta.openai.com/docs/guides/fine-tuning/case-study-i...</a><p>3. More complicated - probably not. It mostly just needs more data with specific training on domain specific areas. Take any field and generate some wrong completions from the data available and some right completions and the output tends to get better. Train it in the opposite fashion and it gets worse.<p>The meat of this comes back to people see ChatGPT as changing its response as an indication of understanding, but it's not. Users are just adding extra constraints on what parts of the language model the model uses to generate text. This is not the same thing.<p>An (perhaps poor) example of that might be where you're deciding to buy a tool that might solve a problem that you have. Evaluating whether it does solve the problem you decide that it doesn't. But then you evaluate whether you can afford the tool based on your budget and you decide you can, so you buy the tool. You haven't changed your mind, you've just constrained the reasons that you're using to evaluate the choice (electronic musicians often refer to this as GAS - Gear Acquisition Syndrome).<p>TL;DR: Stop Anthropomorphizing Language Models
"Truth" is the most problematic concepts in philosophy. The introduction of the concept of the "Truth" undermines truthfulness. (e.g. you can call something "Truth Social")<p>This book<p><a href="https://en.wikipedia.org/wiki/G%C3%B6del,_Escher,_Bach" rel="nofollow">https://en.wikipedia.org/wiki/G%C3%B6del,_Escher,_Bach</a><p>has a set of parables about people trying to paint on a facility to a system very similar to a "truth detector" for GPT-3. The gist of it is that "awareness of truth" makes it possible to make statements like "Am I lying now?"<p>People under GPT-3's spell think that giving correct answers is a minor detail that will be handled in a point revision of it but actually it is a much harder problem than everything they've done so far.
It does acknowledge it's wrong, but it doesn't really understand.<p><a href="https://medium.com/p/cf66b04f3b9f" rel="nofollow">https://medium.com/p/cf66b04f3b9f</a>