<div dir="ltr"><div dir="ltr"><br><div><img src="cid:ii_lhagfovb0" alt="Subjective_CPU.png" width="562" height="276"></div><div><br></div><div>My way of thinking about things agrees with Jason's.  Sense organs (eyes) aren't required for a dreaming system to render the same models into our subjective CPU of consciousness.  And the referent to the word "redness" is to a quality of the model, in the CPU.  There is only an isomporhic modeling relationship between this false colored knowledge of the strawberry, and the much more complex strawberry out there.  So, even though we do know the meaning of "redness" the ability for us to know the meaning of this subjective MODEL of the strawberry out there, is no different from what a large language MODEL can have, by simply having a different isomorphic model composed of only words instead of models composed of subjective properties.</div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Fri, May 5, 2023 at 4:06 AM Jason Resch <<a href="mailto:jasonresch@gmail.com">jasonresch@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="auto"><div><br><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Thu, May 4, 2023, 9:38 PM Gordon Swobe <<a href="mailto:gordon.swobe@gmail.com" target="_blank">gordon.swobe@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="auto">I’m hoping that you, Jason, can see now the difference between what GPT-4 and Bard and I call “simulated understanding,” (what I sometimes also call “understanding” in scare quotes), and true understanding. </div><div dir="auto"><br></div><div dir="auto">Language models have no true understanding of words, or of the world in general. They have only simulated understanding, what I have seen you call the appearance of understanding. There is nobody there to have true understanding.</div><div dir="auto"><br></div><div dir="auto">From this I see you wanting to draw some conclusions about p-zombies, but LLMs do not even remotely resemble humans, zombified or otherwise. They have no bodies or sense organs, to name the most obvious and relevant differences. I think LLMs tell us nothing or next to nothing about p-zombies.</div><div dir="auto"><br></div><div dir="auto">-gts</div><div dir="auto"></div></blockquote></div></div><div dir="auto"><br></div><div dir="auto"><div dir="auto"><br></div><div dir="auto">1. What is required to have a sense organ?</div><div dir="auto"><br></div><div dir="auto">To me it simply requires some conduit of information from the environment. Does a LLM not have this?</div><div dir="auto"><br></div><div dir="auto">Jason </div></div><div dir="auto"><br></div><div dir="auto"><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="auto"><br></div><div dir="auto"><br></div><div dir="auto"><br></div><div dir="auto"><br></div><div dir="auto"><br></div><div dir="auto"><br></div><div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Thu, May 4, 2023 at 6:41 PM Brent Allsop <<a href="mailto:brent.allsop@gmail.com" rel="noreferrer" target="_blank">brent.allsop@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr"><br><div>Hi Gordon,</div><div>Yea, thanks for sending this.</div><div>I agree with  most of this, but think there might be some differences in my views.</div><div>Studying what you are saying has helped me better develop and express my views.</div><div>I'm working on a description, which I'll send on to find out if they are indeed different.</div><div><br></div><div><br></div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Thu, May 4, 2023 at 3:20 PM Gordon Swobe <<a href="mailto:gordon.swobe@gmail.com" rel="noreferrer" target="_blank">gordon.swobe@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr">Brent and Jason,<br><br>I value your insights and intellectual capacity, which is why I'm sharing my thoughts with you both. When I presented my argument on ExI, it seemed to elicit misunderstandings and hostility from others, rather than the intelligent discussion I had hoped for. I could only take so much of it and left.<br><br>Bard is clearly not as well-trained and mature as GPT-4, but it gets my argument. Like humble GPT-4, humble Bard understands there is an important distinction to be made between true understanding and simulated understanding that these virtual friends of ours have no true understanding of words.<br><br>Me: <br>Is this paragraph of mine true?<br><br>To have true understanding of words, one must have some familiarity with the things or ideas they refer to. These things to which words refer are called referents. This familiarity can come from one's own experiences, from one's knowledge of the world, or from one's interactions with others. Referents can be real or imaginary, for example unicorns we read about in a story. Some words refer to other words, but those words also have referents. Ultimately, the referents of all words exist outside of language. <br><br>Bard:<br>Yes, the paragraph is true. The paragraph is about the relationship between words and their referents. It states that in order to have true understanding of words, one must have some familiarity with the things or ideas they refer to. These things to which words refer are called referents. The paragraph goes on to say that this familiarity can come from one's own experiences, from one's knowledge of the world, or from one's interactions with others. It also states that referents can be real or imaginary. Finally, the paragraph concludes by saying that ultimately, the referents of all words exist outside of language.<br><br>Here are some examples to illustrate the paragraph's points:<br><br>The word "dog" refers to a real animal that many people are familiar with.<br>The word "unicorn" refers to an imaginary animal that some people are familiar with because they have read about it in stories.<br>The word "love" refers to a feeling that many people have experienced.<br>The word "democracy" refers to a form of government that many people are familiar with.<br>As the paragraph states, all of these referents exist outside of language. They are not created by language, but rather language is used to refer to them.<br><br>Me:<br>If that paragraph is true and large language models have no access to anything except words and language, how can large language models have true understanding of words? <br><br>Bard:<br>That is a very good question. Large language models are trained on massive amounts of text data, but they do not have any direct experience with the world. They cannot see, hear, smell, taste, or touch anything. They cannot interact with the world in any way. All they have is text.<br><br><b>This means that large language models can only simulate understanding of words.</b> They can learn to associate certain patterns of words with certain referents, but they cannot truly understand what those referents are. They do not have any personal experience with them.<br><br>For example, a large language model might be able to learn to associate the word "dog" with the image of a dog. However, it will never truly understand what a dog is like. It will never know what it feels like to pet a dog, to play with a dog, or to be bitten by a dog.<br><br><b>So, while large language models can be very impressive, it is important to remember that they are only simulating understanding. </b><br><br>-gts</div>
</blockquote></div>
</blockquote></div></div>
</blockquote></div></div></div>
</blockquote></div></div>