As junk web pages written by AI proliferate, the models that rely on that data will suffer.
Good.
“On two occasions I have been asked, 'Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?' I am not able rightly to apprehend the kind of confusion of ideas that could provoke such a question.” - Charles Babbage
The business people adopting AI: "who cares what it's trained on? It's intelligent right? It'll just sort through the garbage and magically come up with the right answers to everything"
Not so hard to imagine given that these people have always seen technical systems as magic.
Of course modern UX design is very much based on getting the right answer with the wrong inputs (autocorrect, etc).
I believe Robustness was the term I learned years ago: the ability of a system to gracefully handle user error, make it easy to recover from or fix, clearly communicate what was wrong etc.
Of course, nothing is ever perfect and humans are very creative at fucking up, and a lot of companies don't seem to take UX too seriously. Particularly when the devs get tunnel vision and forget about user error being a thing....
Garbage in; Garbage out.
Shit-fueled ouroboros
You can't explain it!
Recycle the garbage that comes out... Still more garbage out.
Model degeneration is an already well-known phenomenon. The article already explains well what's going on so I won't go into details, but note how this happens because the model does not understand what it is outputting - it's looking for patterns, not for the meaning conveyed by said patterns.
Frankly at this rate might as well go with a neuro-symbolic approach.
The issue with your assertion is that people don't actually work a similar way. Have you ever met someone who was clearly taught "garbage'?
The issue with your assertion is that people don’t actually work a similar way.
I'm talking about LLMs, not about people.
I know you are, but the argument that an LLM doesn't understand context is incorrect. It's not human level understanding, but it's been demonstrated that they do have a level of understanding.
And to be clear, I'm not talking about consciousness or sapience.
I know you are, but the argument that an LLM doesn’t understand context is incorrect
Emphasis mine. I am talking about the textual output. I am not talking about context.
It’s not human level understanding
Additionally, your obnoxiously insistent comparison between LLMs and human beings boils down to a red herring.
Not wasting my time further with you.
[For others who might be reading this: sorry for the blatantly rude tone but I got little to no patience towards people who distort what others say, like the one above.]
I got little to no patience towards people who distort what others say,
My original reply was meant to be tongue-in-cheek, but I guess I forgot about Poe's law. I'm not a layman, for the record. I've worked with AI for over a decade
Not wasting my time further with you.
Ditto. Have a nice day.
but it's been demonstrated that they do have a level of understanding.
Citation needed
Here you go
A better mathematical system of storing words does not mean the LLM understands any of them. It just has a model that represents the relation between words that it uses.
If I put 10 minus 8 into my calculator I get 2. The calculator doesn't actually understand what 2 means, or what subtracting represents, it just runs the commands that gives the appropriate output.
That's a bad analogy, because the calculator wasn't trained using an artificial neural network literally designed by studying biological brains (aka biological neutral networks).
And "understand" doesn't equate to consciousness or sapience. For example, it is entirely and factually correct to state that an LLM is capable of reasoning. That's not even up for debate. The accuracy of an LLM's reasoning capability is one of the fundamental benchmarks used for evaluating its quality.
But that doesn't mean it's "thinking" in the way most people consider.
Edit: anyone up voting this CileTheSane clown is in the same boat of not comprehending how LLMs work.
it is entirely and factually correct to state that an LLM is capable of reasoning
Citation needed.
If you're going to tell me LLMs are modeled after biological brains and capable of reasoning then I call bullshit on your claims that you actually work in AI.
Imagine you put a man in an enclosed room. There is a slot in the wall where messages get passed through written in Chinese. The man does not speak Chinese or even recognize the written language, he just thinks they're weird symbols.
First the man is shown examples of sequences of symbols to train him. Then he is shown incomplete sequences and asked which symbol comes next. If incorrect he is corrected, if correct he gets cookie. Eventually this man is able to carry on "conversations" with people in Chinese through continued practice.
This man still does not speak Chinese, he is not having reasoned, rational arguments with the people he is conversing with, and if you told him it was a language he's look at you like your crazy. "There's no language here, just if I have these symbols and I next put the one that looks like a man wearing a hat they give me a cookie."
Thinking LLMs are capable of reasoning is the digital equivalent of putting eyes on a pencil then feeling bad when it gets broken in half.
Citation needed.
Certainly!
In machine learning, a neural network (also artificial neural network or neural net, abbreviated ANN or NN) is a model inspired by the structure and function of biological neural networks in animal brains
A neural network is a method in artificial intelligence that teaches computers to process data in a way that is inspired by the human brain.
A neural network is a machine learning program, or model, that makes decisions in a manner similar to the human brain
*A neural network, or artificial neural network, is a type of computing architecture that is based on a model of how a human brain functions *
Would you like some more citations?
Thinking LLMs are capable of reasoning is the digital equivalent of putting eyes on a pencil then feeling bad when it gets broken in half.
In this paper, we present Reasoning via Planning (RAP), a novel LLM reasoning framework that equips LLMs with an ability to reason akin to human-like strategic planning
Source - Reasoning with Language Model is Planning with World Model
Motivated by the observation that adding more concise CoT examples in the prompt can improve LLM reasoning performance
LegalBench - a tool to evaluate the reasoning performance of an LLM in the legal domain.
A paper on benchmarking an LLMs temporal reasoning.
Shall I provide some more?
Wikipedia is not a source.
Amazon is not a source.
Someone trying to sell their LLM to the general public, and therefore simplifying the language to convey a concept is not a source.
These nodes pass data to each other, just like how in a brain, neurons pass electrical impulses to each other.
By that definition my dimmer switch functions like a biological brain because it passes electrical impulses.
In this paper, we present Reasoning via Planning (RAP), a novel LLM reasoning framework that equips LLMs with an ability to reason akin to human-like strategic planning
This prevents LLMs from performing deliber- ate planning akin to human brains,
So does not function like a brain does.
To overcome the limitations, we propose a new LLM reasoning framework
So it's a proposal for a new framework to mimic it, not how LLMs currently function
Aaand I'm going to stop checking your sources now. If you're just going to gish gallop every link from a search page you think agrees with you I'm not going to waste my time reading things you clearly didn't bother to. It took 5 links to get to something that even looks like a source, and it doesn't say what you think it does.
Read your sources and make sure they say what you think they do. If you present me with another pile of links and the first one is invalid I won't bother looking at the 2nd.
AI making itself sick and worthless after flooding the internet with trash just gives me a warm glow.
⢀⣠⣾⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⠀⠀⠀⠀⣠⣤⣶⣶ ⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⠀⠀⠀⢰⣿⣿⣿⣿ ⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣧⣀⣀⣾⣿⣿⣿⣿ ⣿⣿⣿⣿⣿⡏⠉⠛⢿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⡿⣿ ⣿⣿⣿⣿⣿⣿⠀⠀⠀⠈⠛⢿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⠿⠛⠉⠁⠀⣿ ⣿⣿⣿⣿⣿⣿⣧⡀⠀⠀⠀⠀⠙⠿⠿⠿⠻⠿⠿⠟⠿⠛⠉⠀⠀⠀⠀⠀⣸⣿ ⣿⣿⣿⣿⣿⣿⣿⣷⣄⠀⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣴⣿⣿ ⣿⣿⣿⣿⣿⣿⣿⣿⣿⠏⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠠⣴⣿⣿⣿⣿ ⣿⣿⣿⣿⣿⣿⣿⣿⡟⠀⠀⢰⣹⡆⠀⠀⠀⠀⠀⠀⣭⣷⠀⠀⠀⠸⣿⣿⣿⣿ ⣿⣿⣿⣿⣿⣿⣿⣿⠃⠀⠀⠈⠉⠀⠀⠤⠄⠀⠀⠀⠉⠁⠀⠀⠀⠀⢿⣿⣿⣿ ⣿⣿⣿⣿⣿⣿⣿⣿⢾⣿⣷⠀⠀⠀⠀⡠⠤⢄⠀⠀⠀⠠⣿⣿⣷⠀⢸⣿⣿⣿ ⣿⣿⣿⣿⣿⣿⣿⣿⡀⠉⠀⠀⠀⠀⠀⢄⠀⢀⠀⠀⠀⠀⠉⠉⠁⠀⠀⣿⣿⣿ ⣿⣿⣿⣿⣿⣿⣿⣿⣧⠀⠀⠀⠀⠀⠀⠀⠈⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢹⣿⣿ ⣿⣿⣿⣿⣿⣿⣿⣿⣿⠃⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸⣿⣿
the root
Welcome, weary traveler.
How was your journey?
depressing
I'd be very wary of extrapolating too much from this paper.
The past research along these lines found that a mix of synthetic and organic data was better than organic alone, and a caveat for all the research to date is that they are using shitty cheap models where there's a significant performance degrading in the synthetic data as compared to SotA models, where other research has found notable improvements to smaller models from synthetic data from the SotA.
Basically this is only really saying that AI models across multiple types from a year or two ago in capabilities recursively trained with no additional organic data will collapse.
It's not representative of real world or emerging conditions.
provenance requires some way to filter the internet into human-generated and AI-generated content, which hasn’t been cracked yet
It doesn't need to be filtered into human / AI content. It needs to be filtered into good (true) / bad (false) content. Or a "truth score" for each.
We don't teach children to read by just handing them random tweets. We give them books that are made specifically for children. Our filtering mechanism for good / bad content is very robust for humans. Why can't AI just read every piece of "classic literature", famous speeches, popular books, good TV and movie scripts, textbooks, etc?
It doesn’t need to be filtered into human / AI content. It needs to be filtered into good (true) / bad (false) content. Or a “truth score” for each.
That isn't enough because the model isn't able to reason.
I'll give you an example. Suppose that you feed the model with both sentences:
Both sentences are true. And based on vocabulary of both, the model can output the following sentences:
Both are false but the model doesn't "know" it. All that it knows is that "have" is allowed to go after both "cats" and "birds", and that both "feathers" and "fur" are allowed to go after "have".
It's not just a predictive text program. That's been around for decades. That's a common misconception.
As I understand it, it uses statistics from the whole text to create new text. It would be very rare to output "cats have feathers" because that phrase doesn't ever appear in the training data. Both words "have feathers" never follow "cats".
But the fact remains that it doesn't know what a cat or a feather is. All of this is still based purely on statistical frequency and not at all on actual meanings.
and that is exactly how a predictive text algorithm works.
some tokens go in
they are processed by a deterministic, static statistical model, and a set of probabilities (always the same, deterministic, remember?) comes out.
pick the word with the highest probability, add it to your initial string and start over.
if you want variety, add some randomness and don't just always pick the most probable next token.
Coincidentally, this is exactly how llms work. It's a big markov chain, but with a novel lossy compression algorithm on its state transition table. The last point is also the reason why, if anyone says they can fix llm hallucinations, they're lying.
Coincidentally, this is exactly how llms work
Everyone who says this doesn't actually understand how LLMs work.
Multivector word embeddings create emergent relationships that's new knowledge that doesn't exist in the training dataset.
Computerphile did a good video on this well before the LLM craze.
1 - a markov chain only takes previous tokens as input.
2 - It uses a function (in the mathematical sense, so same input results in same output, completely stateless) to generate a set of probabilities for what the next token might be.
3 - The most probable token is picked, else randomness (temperature) is inserted here to choose a different token occasionally.
an llm's internals, the part that's trained is literally the function used in step 2. You could have this function implemented a number of ways, ex you could build a huge table and consult it. Or you could generate it somehow. You could train a big neural network that takes previous tokens as input, and outputs probabilities of tokens as output. You then enumerate its outputs for every possible permutation of inputs and there's your table. This would take too much time and space, so we just run the function on-demand instead. Exact same result. It can be very smart and notice correlations, but ultimately it generates a (virtual) huge static table. This is a completely deterministic process. A trained NN is still a (huge) mathematical function. So the big network that they spend resources training is basically the function used in step 2.
Step 3 is the cause of hallucinations. It's the only nondeterministic part. And it's not part of the llm itself in any way. No matter how smarter the neural network gets, the hallucinations are introduced mainly in step 3. So no, they won't be solving the LLM hallucination problem anytime soon.
because that phrase doesn’t ever appear in the training data.
Eh but LLMs abstract. It has seen " have feathers" and " have fur" quite a lot of times. The problem isn't that LLMs can't reason at all, the problem is that they do employ techniques used in proper reasoning, in particular tracking context throughout the text (cross-attention) but lack techniques necessary for the whole thing, instead relying on confabulation to sound convincing regardless of the BS they spout. Suffices to emulate an Etonian but that's not a high standard.
Workarounds for those sorts of limitations have been developed, though. Chain-of-thought prompting has been around for a while now, and I recall recently seeing an article about a model that had that built right into it; it had been trained to use tags to enclose invisible chunks of its output that would be hidden from the end user but would be used by the AI to work its way through a problem. So if you asked it whether cats had feathers it might respond "Feathers only grow on birds and dinosaurs. Cats are mammals. No, cats don't have feathers." And you'd only see the latter bit. It was a pretty neat approach to improving LLM reasoning.
Both sentences are true. And based on vocabulary of both, the model can output the following sentences:
- Cats have feathers.
- Birds have fur
This is not how the models are trained or work.
Both are false but the model doesn't "know" it. All that it knows is that "have" is allowed to go after both "cats" and "birds", and that both "feathers" and "fur" are allowed to go after "have".
Demonstrably false. This isn't how LLMs are trained or built.
Just considering the contextual relationships between word embeddings that are created during training is evidence enough. Those relationships from the multi-vector fields are an emergent property that doesn't exist in the dataset.
If you want a better understanding of what I just said, take a look at this Computerphile video from four years ago. And this came out before the LLM hype and before ChatGPT 3, which was the big leap in LLMs.
People are already comparing older content with Low Background Steel, as it's uncontaminated
And they're overlooking that radionuclide contamination of steel actually isn't much of a problem any more, since the surge in background radionuclides caused by nuclear testing peaked in 1963 and has since gone down almost back to the original background level again.
I guess it's still a good analogy, though. People bring up Low Background Steel because they think radionuclide contamination is an unsolved problem (despite it having been basically solved), and they bring up "model collapse" because they think it's an unsolved problem (despite it having been basically solved). It's like newspaper stories, everyone sees the big scary front page headline but nobody pays attention to the little block of text retracting it on page 8.
News at 11.
The AI art is inbreeding.
certainly at least a downvote to free will
AI writing, scraped by AI, producing more AI writing...
So not "gray goo" exactly, but "gray slop"?
Water is wet
Is it wet or does it make other things wet?
interdasting
Woah, that was fast.
Maybe we can use it to train the other AIs to help ourselves.
This is a most excellent place for technology news and articles.
