<div dir="ltr">* Bender, not Pender<br><br>"It is really important to distinguish between the word as a sequence of characters as opposed to word in a sense of a pairing of form and meaning." - Bender<br><br>These language models are trained only on the forms of words, the sequences of characters, not on the meanings.<br><br>-gts</div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Sat, Mar 25, 2023 at 3:29 PM Gordon Swobe <<a href="mailto:gordon.swobe@gmail.com">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">I mentioned Emily Pender in another thread. She is Professor of Linguistics and Faculty Director of the Master's Program in Computational Linguistics at University of Washington.<br><br>In the other thread, I made the mistake of introducing her with her Octopus thought experiment which I soon realized from the responses here is easily misinterpreted outside of the context of her general thesis and the academic paper in which she introduced it. <br><br>As I learned from this interview, she and her colleague Koller wrote that paper in response to a twitter debate in which she found herself arguing with non-linguists who insist that language models understand language. Like me, she is critical of such claims. She considers them "hype."<br><br>The relevant material starts at the 26 minute mark.<br><br><a href="https://www.youtube.com/watch?v=VaxNN3YRhBA" target="_blank">https://www.youtube.com/watch?v=VaxNN3YRhBA</a><br> <br></div>
</blockquote></div>