• 5 Posts
Joined 1 year ago
Cake day: June 9th, 2023


  • Sonori@beehaw.orgtoScience Memes@mander.xyzDark Matter
    11 days ago

    To expand on this, we also have mapped it out and know that the amount of dark matter varies wildly between galaxies, with some having basically none while others have far more dark matter than observable matter in them. There’s also a lot of stuff with the early universe that only works if you have something with gravity that doesn’t otherwise interact significantly with matter.

    As Angela Collier puts it, dark matter is not a theory, it is a set of observations.

  • “A computer can never be held accountable, therefore a computer must never make a management decision.”

    Even more importantly when it comes to assessing properly, machine learning, now referred to as AI, has been continuealy shown to not just repeat the biases in its training data, but to significantly exaggerate them.

    Given how significantly and explicitly race has been used to determine and guide so much property and neighborhood development in the training data, I do not look forward to seeing a system that is not only more racist than a post war city council choosing where to build new moterways but which is sold and treated as infallible by the humans operating and litigating it.

    Given the deaths and disaster created by the Horizon Post Office Scandel, I also very much do not look forward to the widespread adoption of software which is inherently and provablly far less accurate, reliable, and auditable than the Horizon software. At least that could only ruin your life if you were a Postmaster and not just any member of the general public who isn’t rich enough to have your affairs handled by a human.

    But hey, on the bright side, if Horizon set UK legal precedent than any person or property agent is fully and unequivocally legally liable for the output of any software they use, after the first few are found guilty for things the procedural text generator they used wrote people might decide its not worth the risk.

  • Generally the term Markov chain is used to discribe a model with a few dozen weights, while the large in large language model refers to having millions or billions of weights, but the fundamental principle of operation is exactly the same, they just differ in scale.

    Word Embeddings are when you associate a mathematical vector to the word as a way of grouping similar words are weighted together, I don’t think that anyone would argue that the general public can even solve a mathematical matrix, much less that they can only comprehend a stool based on going down a row in a matrix to get the mathematical similarity between a stool, a chair, a bench, a floor, and a cat.

    Subtracting vectors from each other can give you a lot of things, but not the actual meaning of the concept represented by a word.

  • To note the obvious, an large language model is by definition at its core a mathematical formula and a massive collection of values from zero to one which when combined give a weighted average of the percentage that word B follows word A crossed with another weighted average word cloud given as the input ‘context’.

    A nuron in machine learning terms is a matrix (ie table) of numbers between zero and 1 by contrast a single human nuron is a biomechanical machine with literally hundreds of trillions of moving parts that darfs any machine humanity has ever built in terms of complexity. This is just a single one of the 86 billion nurons in an average human brain.

    LLM’s and organic brains are completely different and in both design, complexity, and function, and to treat them as closely related much less synonymous betrays a complete lack of understanding of how one or both of them fundamentally functions.

    We do not teach a kindergartner how to write by having them read for thousands of years until they recognize the exact mathematical odds that string of letters B comes after string A, and is followed by string C x percent of the time. Indeed humans don’t naturally compose sentences one word at a time starting from the beginning, instead staring with the key concepts they wish to express and then filling in the phrasing and grammar.

    We also would not expect that increasing from hundreds of years of reading text to thousands would improve things, and the fact that this is the primary way we’ve seen progress in LLMs in the last half decade is yet another example of why animal learning and a word cloud are very different things.

    For us a word actually correlates to a concept of what that word represents. They might make mistakes and missunderstand what concept a given word maps to in a given language, but we do generally expect it to correlate to something. To us a chair is a object made to sit down on, and not just the string of letters that comes after the word the in .0021798 percent of cases weighted against the .0092814 percent of cases related to the collection of strings that are being used as the ‘context’.

    Do I believe there is something intrinsically impossible for a mathematical program to replicate about human thought, probably not. But this this not that, and is nowhere close to that on a fundamental level. It’s comparing apples to airplanes and saying that soon this apple will inevitably take anyone it touches to Paris because their both objects you can touch.

  • Like say, treating a program that shows you the next most likely word to follow the previous one on the internet like it is capable of understanding a sentence beyond this is the most likely string of words to follow the given input on the internet. Boy it sure is a good thing no one would ever do something so brainless as that in the current wave of hype.

    It’s also definitely becuse autocompletes have made massive progress recently, and not just because we’ve fed simpler and simpler transformers more and more data to the point we’ve run out of new text on the internet to feed them. We definitely shouldn’t expect that the field as a whole should be valued what it was say back in 2018, when there were about the same number of practical uses and the foucus was on better programs instead of just throwing more training data at it and calling that progress that will continue to grow rapidly even though the amount of said data is very much finite.

  • Except when it comes to LLM, the fact that the technology fundamentally operates by probabilisticly stringing together the next most likely word to appear in the sentence based on the frequency said words appeared in the training data is a fundamental limitation of the technology.

    So long as a model has no regard for the actual you know, meaning of the word, it definitionally cannot create a truly meaningful sentence. Instead, in order to get a coherent output the system must be fed training data that closely mirrors the context, this is why groups like OpenAi have been met with so much success by simplifying the algorithm, but progressively scrapping more and more of the internet into said systems.

    I would argue that a similar inherent technological limitation also applies to image generation, and until a generative model can both model a four dimensional space and conceptually understand everything it has created in that space a generated image can only be as meaningful as the parts of the work the tens of thousands of people who do those things effortlessly it has regurgitated.

    This is not required to create images that can pass as human made, but it is required to create ones that are truely meaningful on their own merits and not just the merits of the material it was created from, and nothing I have seen said by experts in the field indicates that we have found even a theoretical pathway to get there from here, much less that we are inevitably progressing on that path.

    Mathematical models will almost certainly get closer to mimicking the desired parts of the data they were trained on with further instruction, but it is important to understand that is not a pathway to any actual conceptual understanding of the subject.