Jean Louis <bugs@gnu.support> writes: > * Suhail Singh <suhailsingh...@gmail.com> [2025-03-20 20:12]: >> Jean Louis <bugs@gnu.support> writes: >> >> > I just asked for examples, which I can't find easily. >> >> As someone with a fair amount of experience in the field, it would >> probably do you well to read some relevant research on the subject to >> understand how some of these models work. Regardless, let's see if we >> can address your question. > > Welcome, yet another expert.
Not what I said. >> The question is a little ill-posed, and so some further clarification >> would help. >> >> 1. By large language models, is it fair to assume that you are >> specifically focused _only_ on transformer-based neural networks? > > It is not relevant if I used it, or someone else used it. Not what I was asking. > My humbler question was to show some innovations created by LLMs. Then > it developed to the notion that there are many many examples, but they > can't be easily referenced. As I hope, when I get to the second point below, I'll hopefully see some examples from you regarding what you count as innovation. >> 2. Please provide 10 positive and 10 negative examples of what you mean >> by "innovation". In order for the examples to be of relevance, >> please ensure that the innovative thing is text-based. Otherwise, >> while you may be describing what innovation means to you, it's not >> applicable in the domain where you seek examples of innovation by >> LLMs. For each positive and negative example, please share what >> about the example makes it "innovative" vs not. Please do not use >> any aid from an LLM for the generation of these examples (lest the >> answers dilute your intended meaning of the term). > > While people can strive to change definitions of words, in this case > it is not subjective, rather objective definition. What is or isn't considered "innovation" is not objective. Even if the term may be defined objectively in terms of characteristics such as "new" or "novel", what those terms mean isn't objective. > See: https://en.wikipedia.org/wiki/Innovation as it is well explained there. I didn't request an explanation of the definition. Instead, I requested 10 positive and 10 negative examples (each with their explanation) that demonstrate what the concept means to you. A reference to a Wikipedia article isn't a substitute. Even if the article has examples, it's unclear whether you agree with all of them or not. And even if it has positive examples, it may not have negative examples. And even when it has examples, the reasons why the article deems them to be innovative (or not) may not be why you believe them to be so. And finally, the examples I sought were where the thing being created is text-based. To quote your words: "I just asked for examples, which I can't find easily." > Question is not if human can use some resources to drive or create > innovations. No, but your take on what is or isn't innovation is fair game. And if you're unwilling or unable to share the reasonable number of positive and negative examples I requested, then this conversation is a dead end. > Question is if LLM can do it? Or is it all based on the knowledge > which was already stored inside of the Large Language Model (LLM). Asking whether "blah" can do "it", without confirming what you mean by "blah" (my first clarification), nor providing examples for "it" (my second clarification) is ill-posed. You can keep asking, but it's unsurprising that you've not found answers. And the reason for that isn't necessarily because "blah" can't do "it". > And I kind of think, that your question about "what is my definition > of innovation" was not your own question, I feel you asked maybe > ChatGPT about it. Thank you for sharing your thoughts. It would've been more constructive had you also taken the time to respond to the specific clarifying questions I had asked above. > Thus, it is a pitfall to think that the use of AI will create entirely > new ideas. Regardless of whether that's a pitfall, it certainly is futile to discuss this without clarifying what "AI" means (see clarification 1 that was sought; also, in general, AI is not identical to LLMs), nor providing examples for what "entirely new" means (clarification 2). > After all innovation is about overcoming challenges, isn't it? I wish could tell you, but I'm unable to help since the two clarifications I sought were left unanswered. I'm bowing out of this conversation. -- Suhail --- via emacs-tangents mailing list (https://lists.gnu.org/mailman/listinfo/emacs-tangents)
