Sunday, March 5, 2023

ChatGPT: Our Next Salon Discussion recommended by Ron




Ron suggested this article about artificial intelligence as a topic for our next salon. 

For discussion Friday, March 10.  

Click HERE to read the article as it appears on Henry Kissinger's Website or scroll down to read it here. As time allows, I will update the annotations over the next two days. (They will appear below on this page.) 

ChatGPT Heralds an Intellectual Revolution
By Henry Kissinger,
Eric Schmidt
And Daniel Huttenlocher 

By the way... Henry Kissinger will celebrate his 100th birthday May 27, 2023!









2  "ChatGPT give me 6 references for the treatment of my mother's brain tumor."


3

4





5





6



7


8
9


10
11  Use of quantum theory here ... see discussion note #3 



12



13



14



DISCUSSION
How much of this article was written by Elenor Rhunde...
... by Henry Kissinger?


1. LIST OF Ai FIRST PRINCIPLES from our Salon discussion. (With alacrity: cheerful readiness.)



















2. ChatGPT ... new intellectual revolution? Not until AI understands.
cGPT requires human vetting for accuracy confidence, for example, 

On page 7 (above) ChatGPT was asked to give:

6 references to Kissenger's thoughts on technology.

ChatGPT's response was wrong for five of six listed references.
ChatGPT made up (fabricated) 5 of the six listed items. We know ChatGPT is being "fine tuned" but it has not yet been programmed to deliver reliable citations.  

Would you follow ai's recommendations?  For example: 

"ChatGPT give me 6 references for the treatment of my mother's brain tumor." 

I don't want ChatGPT fabricating results when I'm searching for critical medical information for someone in my family. 

3. Pull out the red flag of skepticism (page 11) whenever you hear QUANTUM THEORY unless it's cited in an article from Cal Tech or MIT or spoken by an actual physicist.


Toby's take:  ChatGPT does not know the accuracy of what it writes. For now it collects data randomly, to create a dialog that writes randomly generated bridge sentences that appear to be human, but are not. For now the first questions is: What percentage of ChapGPT is accurate? Until it is 100% accurate, it should be not be considered ready for mass consumption. Consider  a longitudinal study of a limited beta release.



















FIELDS OF FLOWERS FIELD TRIP MARCH 22, 2024 CARLSBAD, CALIFORNIA

  These stunning photos really pop when viewed full size on your computer. It's true! Click on the photo above when you open this site o...