[ExI] MIT Report Finds That AI Is Leading to “Cognitive Surrender” Among Students

[spike at rainier66.com]

…> On Behalf Of Keith Henson via extropy-chat



 

>…Using AIs to solve real world problems takes common sense that AIs don't really have yet. An example is the SolarSyngas project…Keith

 

 

Keith it isn’t just that.  Current LLMs are good but not yet great, as you discovered.  They don’t know the right answer necessarily, only the answer most commonly published.  You and I have published plenty of material over the decades, here and in other forums.  Those who disagree with us have published even more in some cases.  LLMs don’t know who is right and who is wrong.  It generally takes the majority view.

 

Take some subject that you know one hell of a lot about, more than anyone, then start asking any of the chat bots about it.  Snygas generation is a good example.  Ask the LLMs.  There’s a high probability that you will be not just disappointed but annoyed.

 

This is not to say the LLMs are useless.  They are really good at getting one up to speed on stuff one doesn’t know much about.  They can help Sam Cooke with a science book and the French he took.  It worries me that whenever I query it on something I know really well, it screws it all up, misses the point, doesn’t get it.  But I still use them for learning enough to know diddley or at least as much as Sam knows.

 

spike  

 

 

 

 

 

 

 

 

 

 

 

 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20260919/078e7ae0/attachment.htm>