[kwlug-disc] ChatGTP + Bing: we're entering a new paradigm and it's mind-blowing

[Khalid Baheyeldin]

There is a huge gap between reality and expectation ...

The one thing I see that is a huge leap is the user interaction with the
AI: that is where the innovation has progressed.
Understanding (mostly) what you type, and doing searches on its data set
(which is mainly data from crawling the web), and providing responses using
its "AI" that is trained on that set.

Those who are clamouring about AI's capabilities are non-specialists
(mainly tech and non-tech journalists, Youtubers, ...etc.)

Here are two videos by two professors, one in physics and one in
astophysics.

https://youtu.be/GBtfwa-Fexc

https://youtu.be/K0cmmKPklp4

There are real basic mistakes, like 5 is larger than 7.
Yet, the AI is stubborn and holds its position for a while.

Here is the underlying issue:

ChatGPT and other "FOMO" AI services rely on information that is crawled
from the web.
We all know that the quality and accuracy of the information (as a whole)
is lacking, varying from inaccurate to outright wrong.

Google launched its internal Bard AI, and it got an answer wrong in its ad.

The real world result is that Google's market value tumbled, on the
assumption that Google is subpar to ChatGPT, while the underlying training
data is the same.

The age old GIGO (Garbage In Garbage Out) remains true

https://www.npr.org/2023/02/09/1155650909/google-chatbot--error-bard-shares

Will this improve? Not without refining the training data set.
Maybe in favour of one that is curated and fact-checked?

Remains to be seen how it will handle polarizing domains, such as political
and religious topics?
Even on Wikipedia, such topics cannot be settled to the satisfaction of all
concerned.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://kwlug.org/pipermail/kwlug-disc_kwlug.org/attachments/20230215/16cd087c/attachment.htm>