John It seems to me that an emphasis on ’semantics rather than syntax’ sets up an analytic frame focusing only on entities operating within a mode of Secondness - and ignoring the mode of Thirdness operating in syntax..
Edwina > On Mar 11, 2024, at 3:27 PM, John F Sowa <s...@bestweb.net> wrote: > > In my previous note, I forgot to include a link to the updated (March 8} > slides for my talk on March 6. Here is the URL: > https://ontologforum.s3.amazonaws.com/OntologySummit2024/TrackA/LLMs-are-clueless--JohnSowa_20240228.pdf > . > > I also received an offline note about a linguistic theory that emphasizes > semantics rather than syntax: > The method of Generative semantics by Seuren, > https://www.mpi.nl/sites/default/files/2020-07/Seuren_Abralin_Article_2020.pdf > . Other linguists and computational linguists have proposed, developed, > and/or implemented related versions. > > Methods that emphasize semantics have been used in conjunction with ontology > to correct and avoid the errors and hallucinations created by LLMs. For > critical applications, 99% correct can be a disaster. Nobody wants to fly > in an airplane that has a 1% chance of crashing. > > LLMs are very good for translating linear languages and notations. But when > accuracy is essential, precise semantics is much more important than elegant > syntax. > > I also want to emphasize Section 3. That begins with slide 32, which has the > title Neuro-Cognitive Cycles. The word 'cognitive' is much more general than > 'symbolic', since it includes images as well as linear notations for > language. Note slide 7, which shows an image in the mind of a policeman, and > the attempt by a man who is trying to reconstruct an image from a verbal > explanation. > > In slide 24, I added a picture of a baby who is using sign language. For > multi-dimensional topics, a sign language can be more detailed and precise > that a spoken language. > > This section also emphasizes Peirce's methods of reasoning in Slides 33 and > 34, and their applications in the remaining slides. Slide 35 on the Central > Executive, as defined by neuroscientists, shows how to avoid the errors, > hallucinations, and dangers created by the Large Language Models (LLMs): > Include a Central Executive, which has the responsibility and the power to > evaluate any proposed language or actions and revise or reject those that may > be erroneous or even dangerous. > > Also note slide 39 on "Wicked Problems"; slide 40, which explains "Why Humans > are not obsolete; and Slide 41, which asks whether there is "A Path to AGI?" > The answer to that question is joke by George Burns, which might be taken > seriously. > > That reminds me of a remark by Ludwig Wittgenstein: "It's possible to write > a book on philosophy that consists entirely of jokes." A Zen Buddhist could > write a book on religion that consists entirely of jokes. Depending on the > definition of 'joke', somebody might say that they have. > > John > _______________________________________________ > CG mailing list -- c...@lists.iccs-conference.org > To unsubscribe send an email to cg-le...@lists.iccs-conference.org
_ _ _ _ _ _ _ _ _ _ ARISBE: THE PEIRCE GATEWAY is now at https://cspeirce.com and, just as well, at https://www.cspeirce.com . It'll take a while to repair / update all the links! ► PEIRCE-L subscribers: Click on "Reply List" or "Reply All" to REPLY ON PEIRCE-L to this message. PEIRCE-L posts should go to peirce-L@list.iupui.edu . ► To UNSUBSCRIBE, send a message NOT to PEIRCE-L but to l...@list.iupui.edu with UNSUBSCRIBE PEIRCE-L in the SUBJECT LINE of the message and nothing in the body. More at https://list.iupui.edu/sympa/help/user-signoff.html . ► PEIRCE-L is owned by THE PEIRCE GROUP; moderated by Gary Richmond; and co-managed by him and Ben Udell.