[ExI] ChatGPT 'Not Interesting' for creative works

[spike at rainier66.com]

…> On Behalf Of Adrian Tymes via extropy-chat

…

 

>…Side thought: let's say something like ChatGPT gets trained on a bunch of classified/secret material, then gets used in public.  Given the reliability or lack thereof of its adherence to the facts, and thus the reasonable questions as to whether its output truthfully corresponds to what is in said material, would its output necessarily be classified? Adrian

 

 

 

For ChatGPT to have access to that material, it would need to be cleared.  Since it cannot be persuaded of negative consequences of revealing classified documentation, the output of the software itself would need to become classified.

 

spike

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20230307/8851b536/attachment.htm>