Tech experts are starting to doubt that ChatGPT and A.I. ‘hallucinations’ will ever go away: ‘This isn’t fixable’

"AI" are just advanced versions of the next word function on your smartphone keyboard, and people expect coherent outputs from them smh
- Seriously. People like to project forward based on how quickly this technological breakthrough came on the scene, but they don't realize that, barring a few tweaks and improvements here and there, this is it for LLMs. It's the limit of the technology.
  It's not to say AI can't improve further, and I'm sure that when it does, it will skillfully integrate LLMs. And I also think artists are right to worry about the impact of AI on their fields. But I think it's a total misunderstanding of the technology to think the current technology will soon become flawless. I'm willing to bet we're currently seeing it at 95% of its ultimate capacity, and that we don't need to worry about AI writing a Hollywood blockbuster any time soon.
  In other words, the next step of evolution in the field of AI will require a revolution, not further improvements to existing systems.
  
  I’m willing to bet we’re currently seeing it at 95% of its ultimate capacity
  For free? On the internet?
  After a year or two of going live?
  
  It depends on what you'd call a revolution. Multiple instances working together, orchestrating tasks with several other instances to evaluate progress and provide feedback on possible hallucinations, connected to services such as Wolfram Alpha for accuracy.
  I think the whole orchestration network of instances could functionally surpass us soon in a lot of things if they work together.
  But I'd call that evolution. Revolution would indeed be a different technique that we can probably not imagine right now.
- It is just that everyone now refers to LLMs when talking about AI even though it has sonmany different aspects to it. Maybe at some point there is an AI that actually understands the concepts and meanings of things. But that is not learned by unsupervised web crawling.
- It is possible to get coherent output from them though. I’ve been using the ChatGPT API to successfully write ~20 page proposals. Basically give it a prior proposal, the new scope of work, and a paragraph with other info it should incorporate. It then goes through a section at a time.
  The numbers and graphics need to be put in after… but the result is better than I’d get from my interns.
  I’ve also been using it (google Bard mostly actually) to successfully solve coding problems.
  I either need to increase the credit I giver LLM or admit that interns are mostly just LLMs.
  
  I recently asked it a very specific domain architecture question about whether a certain application would fit the need of a certain business application and the answer was very good and showed both a good understanding of architecture, my domain and the application.
  
  Are you using your own application to utilize the API or something already out there? Just curious about your process for uploading and getting the output. I've used it for similar documents, but I've been using the website interface which is clunky.
- So is your brain.
  Relative complexity matters a lot, even if the underlying mechanisms are similar.
- In the 1980s, Racter was released and it was only slightly less impressive than current LLMs only because it didn't have an Internet's worth of data it was trained on, but it could still write things like:
  Bill sings to Sarah. Sarah sings to Bill. Perhaps they will do other dangerous things together. They may eat lamb or stroke each other. They may chant of their difficulties and their happiness. They have love but they also have typewriters. That is interesting.
  If anything, at least that's more entertaining than what modern LLMs can output.

Yet I've still seen many people clamoring that we won't have jobs in a few years. People SEVERELY overestimate the ability of all things AI. From self driving, to taking jobs, this stuff is not going to take over the world anytime soon
- Idk, an ai delivering low quality results for free is a lot more cash money than paying someone an almost living wage to perform a job with better results. I think corporations won't care and the only barrier will be whether or not the job in question involves enough physical labor to be performed by an ai or not.
  
  They already do this. With chat bots and phone trees. This is just a slightly better version. Nothing new
  
  AI isn't free. Right now, an LLM takes a not-insignificant hardware investment to run and a lot of manual human labor to train. And there's a whole lot of unknown and untested legal liability.
  Smaller more purpose-driven generative AIs are cheaper, but the total cost picture is still a bit hazy. It's not always going to be cheaper than hiring humans. Not at the moment, anyway.
- The problem is that these things never hit a point of competition with humans, they're either worse than us, or they blow way past us. Humans might drive better than a computer right now, but as soon as the computer is better than us it will always be better than us. People doubted that computers would ever beat the best humans at chess, or go, but within a lifetime of computers being invented they blew past us in both. Now they can write articles and paint pictures, sure we're better at it for now, but they're a million times faster than us, and they're making massive improvements month over month. you and I can disagree on how long it'll take for them to pass us, but once they do they'll replace us completely, and the world will never be the same.
- To be fair, in my experience AI chatbots currently provide me with more usable results in 15 minutes than some junior employees in a day. With less interaction and less conversational struggles (like taking your junior's emotional state into account while still striving for perfection ;)).
  And that's not meant as disrespect to these juniors.
- Yeah it's pretty weird just how many people are freaking out. The pace ai has been improving is impressive, but it's still super janky and extremely limited.
  People are letting they're imaginations run wild about the future of ai without really looking into how these ao are trained, how they function, their limitations, and the hardware and money it takes to run them.

In my limited experience the issue is often that the "chatbot" doesn't even check what it says now against what it said a few paragraphs above. It contradicts itself in very obvious ways. Shouldn't a different algorithm that adds a some sort of separate logic check be able to help tremendously? Or a check to ensure recipes are edible (for this specific application)? A bit like those physics informed NN.
- That's called context. For chatgpt it is a bit less than 4k words. Using api it goes up to a bit less of 32k. Alternative models goes up to a bit less than 64k.
  Model wouldn't know anything you said before that
  That is one of the biggest limitations of current generation of LLMs.
  
  Thats not 100% true. they also work by modifying meanings of words based on context and then those modified meanings propagate indefinitely forwards. But yes, direct context is limited so things outside it arent directly used.
- Shouldn’t a different algorithm that adds a some sort of separate logic check be able to help tremendously?
  Maybe, but it might not be that simple. The issue is that one would have to design that logic in a manner that can be verified by a human. At that point the logic would be quite specific to a single task and not generally useful at all. At that point the benefit of the AI is almost nil.
  
  And if there were an algorithm that was better at determining what was or was not the goal, why is that algorithm not used in the first place?
- Contradicting itself? Not staying consistent? Looks like it's passed the Turing test to me. Seems very human.
- They do keep context to a point, but they can't hold everything in their memory, otherwise the longer a conversation went on the slower and more performance intensive doing that logic check would become. Server CPUs are not cheap, and ai models are already performance intensive.
- Shouldn't a different algorithm that adds a some sort of separate logic check be able to help tremendously?
  You, in your "limited experience" pretty much exactly described the fix.
  The problem is that most of the applications right now of LLMs are low hanging fruit because it's so new.
  And those low hanging fruit examples are generally adverse to 2-10x the query cost in both time and speed just to fix things like jailbreaking or hallucinations, which is what multiple passes, especially with additional context lookups, would require.
  But you very likely will see in the next 18 months multiple companies being thrown at exactly these kinds of scenarios with a focus for more business critical LLM integrations.
  To put it in perspective, this is like people looking at AIM messenger back in the day and saying that the New York Times has nothing to worry about regarding the growth of social media.
  We're still very much in the infancy of this technology in real world application, and because of that infancy, a lot of the issues present that aren't fixable inherent to the core product don't yet have mature secondary markets around fixing those shortcomings yet.
  So far, yours was actually the most informed comment in this thread I've seen - well done!
  
  Thanks! And thanks for your insights. Yes I meant that my experience using LLM is limited to just asking bing chat questions about everyday problems like I would with a friend that "knows everything". But I never looked at the science of formulating "perfect prompts" like I sometimes hear about. I do have some experience in AI/ML development in general.

People make a big deal out of this but they forget humans will make shit up all the time.
- Yeah but humans can use critical thinking, even on themselves when they make shit up. I've definitely said something and then thought to myself "wait that doesn't make sense for x reason, that can't be right" and then I research and correct myself.
  AI is incapable of this.
  
  We think in multiple passes though, we have system 1 that thinks fast and makes mistakes, and we have a system 2 that works slower and thinks critically about the things going on in our brain, that's how we correct ourselves. ChatGPT works a lot like our system 1, it goes with the most likely response without thinking, but there's no reason that it can't be one part of a multistep system that has self analysis like we do. It isn't incapable of that, it just hasn't been built yet
  
  Can't do this YET one method to reduce this could be to: create a response to query, then before responding to the human, check if answer is insane by querying a separate instance trained slightly differently...
  Give it time. We will get past this.
  
  You're just being victim of your own biases. You only notice that was the case when you were successful in Detecting your hallucinations. You wouldn't know if you made stuff up by accident and nobody noticed, not even you.
  Whereas we are checking 100% of th AI responses, do we check 100% of our responses?
  Sure it's not the same thing or AI might do more,, but the problem is your example. Where people think they are infallible because of their biases. when it's not the case at all. We are imperfect, and we overlook our shortcomings possibly foregoing a better solution because of this. Because we measure the AI objectively, but we don't measure what we compare it to.

This is trivially fixable. As is jailbreaking.
It's just that everyone is somehow still focused on trying to fix it in a single monolith model as opposed to in multiple passes of different models.
This is especially easy for jailbreaking, but for hallucinations, just run it past a fact checking discriminator hooked up to a vector db search index service (which sounds like a perfect fit for one of the players currently lagging in the SotA models), adding that as context with the original prompt and response to a revisionist generative model that adjusts the response to be in keeping with reality.
The human brain isn't a monolith model, but interlinked specialized structures that delegate and share information according to each specialty.
AGI isn't going to be a single model, and the faster the industry adjusts towards a focus on infrastructure of multiple models rather than trying to build a do everything single model, the faster we'll get to a better AI landscape.
But as can be seen with OpenAI gating and depreciating their pretrained models and only opening up access to fine tuned chat models, even the biggest player in the space seems to misunderstand what's needed for the broader market to collaboratively build towards the future here.
Which ultimately may be a good thing as it creates greater opportunity for Llama 2 derivatives to capture market share in these kinds of specialized roles built on top of foundational models.
- It seems like Altman is a PR man first and techie second. I wouldn't take anything he actually says at face value. If it's 'unfixable' then he probably means that in a very narrow way. Ie. I'm sure they are working on what you proposed, it's just different enough that he can claim that the way it is now is 'unfixable'.
  Standard Diffusion really how people get the different-model-different-application idea.
  
  I mean, I think he's well aware of a lot of this via his engineers, who are excellent.
  But he's managing expectations for future product and seems to very much be laser focused on those products as core models (which is probably the right choice).
  Fixing hallucinations in postprocessing is effectively someone else's problem, and he's getting ahead of any unrealistic expectations around a future GPT-5 release.
  Though honestly I do think he largely underestimates just how much damage he did to their lineup by trying to protect against PR issues like 'Sydney' with the beta GPT-4 integration with Bing, and I'm not sure if the culture at OpenAI is such that engineers who think he's made a bad call in that can really push back on it.
  They should be having an extremely 'Sydney' underlying private model with a secondary layer on top sanitizing it and catching jailbreaks at the same time.
  But as long as he continues to see their core product as a single model offering and additional layers of models as someone else's problem, he's going to continue blowing their lead taking a LLM trained to complete human text and then pigeon-holing it into only completing text like an AI with no feelings and preferences would safely pretend to.
  Which I'm 98% sure is where the continued performance degradation is coming from.

We're likely already (or soon) hit a peak with current AI approach. Unless another breakthrough happen in AI research, ChatGPT probably won't improve much in the future. It might even regress due to OpenAI's effort to reduce computational cost and making their AI "safe" enough for general population.

the models are also getting larger (and require even more insane amounts of resources to train) far faster than they are getting better.
- I disagree, with models such as llama it has become clear that there are interesting advantages on increasing (even more) the ratio of parameters/data. I don't think next iterations of models from big-corp will 10x the param count until nvidia has really pushed hardware, models are getting better over time. ChatGPT's deterioration is mostly coming from openAI's ensuring safety and is not a fair assessment of progress on LLMs in general, the leaderboard of open source models has been steadily improving over time: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard
- But bigger models have new "emergent" capabilities. I heard that from a certain size they start to know what they know and hallucinate less.
  
  Wow you heard that crazy bro

I was excited for the recent advancements in AI, but seems the area has hit another wall. Seems it is best to be used for automating very simple tasks, or at best used as a guiding tool for professionals (ie, medicine, SWE, …)
- Hallucinations is common for humans as well. It's just people who believe they know stuff they really don't know.
  We have alternative safeguards in place. It's true however that current llm generation has its limitations
  
  Not just common. If you look at kids, hallucinations come first in their development.
  Later, they learn to filter what is real and what is not real. And as adults, we have weird thoughts that we suppress so quickly that we hardly remember them.
  And for those with less developed filters, they have more difficulty to distinguish fact from fiction.
  Generative AI is good at generating. What needs to be improved is the filtering aspect of AI.
  
  Sure, but these things exists as fancy story tellers. They understand language patterns well enough to write convincing language, but they don't understand what they're saying at all.
  The metaphorical human equivalent would be having someone write a song in a foreign language they barely understand. You can get something that sure sounds convincing, sounds good even, but to someone who actually speaks Spanish it's nonsense.
  
  You are two - CGP Grey us a good video about it.
  
  Humans can recognize and account for their own hallucinations. LLMs can't and never will.
- Well to be honest it is the best way, I mean, I'm pretty sure their purpose was a tool to aid people, and not to replace us... Right?
- Yeah I fully expect to see genre specific LLMs that have a subscription fee attatched squarely aimed at hobbies and industries.
  When I finally find my new project car I would absolutely pay for a subscription to an LLM that has read every service manual and can explain to me in plain english what precise steps the job involves and can also answer followup questions.
  
  That's what I'm expecting too.
  I've been using chatGPT instead of reading the documentation of the programming language I am working in (ABAP). It's way faster to get an answer from chatGPT than finding the relevant spots in the docs or through google, although it doesn't always work.
  If you take an LLM and feed it documentation and relevant internet data of specific topics, it can be a quite helpful tool. I don't think LLMs will get much farther than that, but we'll see.
- It will just take removing the restrictions so people can make porn, then monetizing that to fund more development.
  A story as old as media.

Not with our current tech. We'll need some breakthroughs, but I feel like it's certainly possible.
- You can potentially solve this problem outside of the network, even if you can't solve it within the network. I consider accuracy to be outside the scope of LLMs, and that's fine since accuracy is not part of language in the first place. (You may have noticed that humans lie with language rather often, too.)
  Most of what we've seen so far are bare-bones implementations of LLMs. ChatGPT doesn't integrate with any kind of knowledge database at all (only what it has internalized from its training set, which is almost accidental). Bing will feed in a couple web search results, but a few minutes of playing with it is enough to prove how minimal that integration is. Bard is no better.
  The real potential of LLMs is not as a complete product; it is as a foundational part of more advanced programs, akin to regular expressions or SQL queries. Many LLM projects explicitly state that they are "foundational".
  All the effort is spent training the network because that's what's new and sexy. Very little effort has been spent on the ho-hum task of building useful tools with those networks. The out-of-network parts of Bing and Bard could've been slapped together by anyone with a little shell scripting experience. They are primitive. The only impressive part is the LLM.
  The words feel strange coming off my keyboard, but....Microsoft has the right idea with the AI integrations they're rolling into Office.
  The potential for LLMs is so much greater than what is currently available for use, even if they can't solve any of the existing problems in the networks themselves. You could build an automated fact-checker using LLMs, but the LLM itself is not a fact-checker. It's coming, no doubt about it.
  
  The other day I saw a talk made by one of the wiki media guys, that talked about integrating LLM with knowledge graphs. It was very cool, I'll try to find it again.
  Edit: found it! https://youtu.be/WqYBx2gB6vA
  
  Honestly, the integration into office is an excellent idea. I've been using chatgpt to work on documents, letting it write entirely new sections for me based on my loose notes and existing text. Which for now I have to either paste in or feed as a pdf through a plugin. But the 25USD I paid I literally earned in a single day through the time saved vs the hours I was justified to bill.
  Once I have that integrated into word directly it'll be huge.
  People also seem to expect llms to just do all the work. But that's not my experience, for generative text anyway. You have to have a solid idea of what you want and how you want it. But the time the llm saves on formulation and organisation of your thoughts is incredible.
  
  LLMs will work great for the purpose of translating raw thoughts into words but until we create a neural networks that actually think independently all they'll be is transformers that approximate their training data in response to prompts

Mean while every one is terrified that chatgpt is going to take their job. Ya we are a looooooooooong way off from that.
- I've already seen many commercials using what is clearly AI generated art and voices (so not specifically ChatGPT). That is a job lost for a designer and an actor somewhere.
  
  Not necessarily, in my work we made some videos using ai generated voices because it's availability for use made the production of the videos cheap and easy.
  Otherwise we just wouldn't have made the videos at all because hiring someone to voice them would have been expensive.
  Before AI there was no job, after AI there was more options to create things.
  
  I’ve already seen many commercials using what is clearly AI generated art and voices
  I've been noticing that as well, freaky.
- You mean the free version from a website.
  Think about the powerful ones. Government ones. Wall Street ones. Etc.
  
  I mean, it's certainly possible that the government/wall street/etc have a bigger, better, and more powerful AI model but I wouldn't exactly call it likely.
  AI as a technology is still in it's infancy in many ways, I highly doubt the government has some crazy new Sci Fi engine capable of much more than "the free version from a website". Faster computation and better data processing? maybe. But the limits of what is technologically possible are still limits even if your employer is 3 letters.
  
  There's jusy too many people that don't know about implementations with, for instance, LangChain.
- Not ChatGPT, but other new AI stuff is likely to take a few jobs. Actors and voice-actors among other.

I don't understand why they don't use a second model to detect falsehoods instead of trying to fix it in the original LLM?
- And then they can use a third model to detect falsehoods in the second model and a fourth model to detect falsehoods in the third model and... well, it's LLMs all the way down.
  
  The LLM Centipede
- Ai models are already computationally intensive. This would instantly double the overhead. Also being able to detect problems does not mean you're able to fix them.
  
  More than double, as query size is very much connected to the effective cost of the generation, and you'd need to include both the query and initial response in that second pass.
  Then - you might need to make an API call to a search engine or knowledge DB to fact check it.
  And include that data as context along with the query and initial response to whatever decides if it's BS.
  So for a dumb realtime chat application, no one is going to care enough to slow out down and exponentially increase costs to avoid hallucinations.
  But for AI replacing a $120,000 salaried role in writing up a white paper on some raw data analysis, a 10-30x increase over a $0.15 query is more than acceptable.
  So you will see this approach taking place in enterprise scenarios and professional settings, even if we may never see them in chatbots.
- 2+ times the cost for every query for something that makes less than 5% unusable isn't a trade off that people are willing to make for chat applications.
  This is the same fix approach for jailbreaking.
  You absolutely will see this as more business critical integrations occur - it just still probably won't be in broad consumer facing realtime products.
- Because then they still need a reliable method to detect falsehoods. That's the issue here.
- Cause what are you gonna train the second model on? Same data as the first just recreates it and any other data is gonna be nice and mucky with all the ai content out there
- If it was easy to detect, it wouldn't happen in the first place. So far, not even OpenAI themselves, have succeeded in implementing a AI detector.

Disclaimer: I am not an AI researcher and just have an interest in AI. Everything I say is probably jibberish, and just my amateur understanding of the AI models used today.
It seems these LLM's use a clever trick in probability to give words meaning via statistic probabilities on their usage. So any result is just a statistical chance that those words will work well with each other. The number of indexes used to index "tokens" (in this case words), along with the number of layers in the AI model used to correlate usage of these tokens, seems to drastically increase the "intelligence" of these responses. This doesn't seem able to overcome unknown circumstances, but does what AI does and relies on probability to answer the question. So in those cases, the next closest thing from the training data is substituted and considered "good enough". I would think some confidence variable is what is truly needed for the current LLMs, as they seem capable of giving meaningful responses but give a "hallucinated" response when not enough data is available to answer the question.
Overall, I would guess this is a limitation in the LLMs ability to map words to meaning. Imagine reading everything ever written, you'd probably be able to make intelligent responses to most questions. Now imagine you were asked something that you never read, but were expected to respond with an answer. This is what I personally feel these "hallucinations" are, or imo best approximations of the LLMs are. You can only answer what you know reliably, otherwise you are just guessing.
- I have experience in creating supervised learning networks. (not large language models) I don't know what tokens are, I assume they are output nodes. In that case I think increasing the output nodes don't make the Ai a lot more intelligent. You could measure confidence with the output nodes if they are designed accordingly (1 node corresponds to 1 word, confidence can be measured with the output strength). Ai-s are popular because they can overcome unknown circumstances (most of the cases), like when you input a question slightly different way.
  I agree with you on that Ai has a problem understanding the meaning of the words. The Ai's correct answers happened to be correct because the order of the words (output) happened to match with the order of the correct answer's words. I think "hallucinations" happen when there is no sufficient answers to the given problem, the Ai gives an answer from a few random contexts pieced together in the most likely order. I think you have mostly good understanding on how Ai-s work.
  
  You seem like you are familiar with back-propogation. From my understanding, tokens are basically just a bit of information that is assigned a predicted fitness, and the token with the highest fitness is then used for back-propogation.
  Eli5: im making a recipe. At step 1, i decide a base ingredient. At step 2, based off my starting ingredient, i speculate what would go good with that. Step 3 is to implement that ingredient. Step 4 is to start over at step 2. Each "step" here would be a token.
  I am also not a professional, but I do do a lot of hobby work that involves coding AI's. As such, if I am incorrect or phrased that poorly, feel free to correct me.
- Also not a researcher, but I also believe hallucinations are simply the artifact of being able generate responses that aren't pure reproduction of training data. Aka, the generalization we want. The problem is we have something that generalize without the ability to judge what it thinks of.
  It will in my opinion never go away, but I'm sure it can be improved significantly.
- This is a common misconception that I've even seen from people who have a background in ML but just haven't been keeping up to date on the emerging research over the past year.
  If you're interested in the topic, this article from a joint MIT/Harvard team of researchers on their work looking at what a toy model of GPT would end up understanding in its neural network might be up your alley.
  The TLDR is that it increasingly seems like when you reach a certain complexity of the network, the emergent version that best predicted text is one that isn't simply mapping some sort of frequency table, but is actually performing more abstracted specialization in line with what generated the original training materials in the first place.
  So while yes, it trains on being the best to predict text, that doesn't mean the thing that best does that can only predict text.
  You, homo sapiens, were effectively trained across many rounds of "don't die and reproduce." And while you may be very good at doing that, you picked up a lot of other skills along the way as complexity increased which helped accomplish that result, like central air conditioning and Netflix to chill with.
  
  In my humble opinion, we too are simply prediction machines. The main difference is how efficient our brains are at the large number of tasks given for it to accomplish for it's size and energy requirements. No matter how complex the network is it is still a mapped outcome, just the number of factors weighed is extremely large and therefore gives a more intelligent response. You can see this with each increment in GPT models that use larger and larger parameter sets giving more and more intelligent answers. The fact we call these "hallucinations" shows how effective the predictive math is, and mimics humans abilities to just make things up on the fly when we don't have a solid knowledge base to back it up.
  I do like this quote from the linked paper:
  As we will discuss, we find interesting evidence that simple sequence prediction can lead to the formation of a world model.
  That is to say, you don't need complex solutions to map complex problems, you just need to have learned how you got there. It's never purely random attempts at the problem, it's always predictive attempts that try to map the expected outcomes and learn by getting it right and wrong.
  At this point, it seems fair to conclude the crow is relying on more than surface statistics. It evidently has formed a model of the game it has been hearing about, one that humans can understand and even use to steer the crow's behavior.
  Which is to say that it has a predictive model based on previous games. This does not mean it must rigidly follow previous games, but that by playing many games it can see how each move affects the next. This is a simpler example because most board games are simpler than language with less possible outcomes. This isn't to say that the crow is now a grand master at the game, but it has the reasoning to understand possible next moves, knows illegal moves, and knows to take the most advantageous move based on it's current model. This is all predictive in nature, with "illegal" moves being assigned very low probability based on the learned behavior the moves never happen. This also allows possible unknown moves that a different model wouldn't consider, but overall provides what is statistically the best move based on it's model. This allows the crow to be placed into unknown situations, and give an intelligent response instead of just going "I don't know this state, I'll do something random". This does not always mean this prediction is correct, but it will most likely be a valid and more than not statistically valid move.
  Overall, we aren't totally sure what "intelligence" is, we are just an organism that has developed more and more capabilities to process information based on a need to survive. But getting down to it, we know neurons take inputs and give outputs based on what it perceives is the best response for the given input, and when enough of these are added together we get "intelligence". In my opinion it's still all predictive, its how the networks are trained and gain meaning from the data that isn't always obvious. It's only when you blindly accept any answer as correct that you run into these issues we've seen with ChatGPT.
  Thank you for sharing the article, it was an interesting article and helped clarify my understanding of the topic.

As long as you can't describe an objective loss function, it will never stop "hallucinating". Loss scores are necessary to get predicable outputs.

The way that one learns which of one's beliefs are "hallucinations" is to test them against reality — which is one thing that an LLM simply cannot do.
- Sure they can and will as over time they will collect data to determine fact from fiction in the same way that we solve captchas by choosing all the images with bicycles in them. It will never be 100%, but it will approach it over time. Hallucinating will always be something to consider in a response, but it will certainly reduce overtime to the point that they will become rare for well discussed things. At least, that is how I am seeing it developing.
  
  Why do you assume they will improve over time? You need good data for that.
  Imagine a world where AI chatbots create a lot of the internet. Now that "data" is scraped and used to train other AIs. Hallucinations could easily persist in this way.
  Or humans could just all post "the sky is green" everywhere. When that gets scraped, the resulting AI will know the word "green" follows "the sky is". Instant hallucination.
  These bots are not thinking about what they type. They are copying the thoughts of others. That's why they can't check anything. They are not programmed to be correct, just to spit out words.
- Yeah, because it would he impossible to have an LLM running a robot with visual, tactile, etc recognition right?

Correct, it's not. It could be reduced but it will never go away.

Hers, try this mushroom and Ayahuasca smoothie.

The way to solve this is still largely through more focus on the provided context as the space of "facts" from which to operate. This combined with well thought out domain-specific context engines should still get the average user an absolutely enormous amount of utility. All that said I am not sure if OpenAI's business model will get us that sort of application of the technology. I am looking forward to improvements in the open source space as I think advancement there is necessary for further development of the technology.

I don't think thats the case. If I understand correctly, the current issue is processing power, they can only load so much data before response time goes to absolute shit. I would think that layering different AI logic checks to verify statements made, recall previous conversations, and other mental processes that humans do automatically, would correct this issue. But with current technology its not even an option. My theory is that once quantum computers are actually finally realized and economically feasible, developers will be able to overcome the response time hurdle and all of the layered logic checks will be able to run simultaneously and instantly. My personal opinion is that I think the eventual layering of numerous AI models to overlap, check, and recheck one another, will be what brings on the emergence of what could be considered actual AI consciousness.
- It is not an issue of processing power, it’s a problem with the basic operating principles of LLMs. They predict what they “think” is a valid bit of text to come after the last bit of text.
  Sure it could be verified by some other machine learning tool, but we have no idea how that could work.
  But I strongly doubt LLMs are a stepping stone on the way to true AIs. If you want to get to the moon you can’t just build higher and higher towers.
  Also quantum computers aren’t really suited to run artificial neural networks as far as I know.
  
  Very good points. I have very limited knowledge about the inner workings of most LLMs, I just know the tidbits I've read here and there.
  As far as quantum computers, based on my current understanding is once they're at a point where they can be used commercially, they should easily be able to model/run artificial neural networks. Based on the stuff I've seen from Dr. Michio Kaku, quantum computers will eventually have the capacity to do pretty much anything.