JUST PRECAUTION

 Suman ,

 

In case you are training Modi AI , with Modi’s past speeches as INPUTS , then I am wondering about following scenario

 

In hundreds of his speeches , he must have used thousands of WORDS

 

Based on the WORDS contained in the questioner’s question , AI will try and put together an answer , incorporating some ( same or other ) words , in a sequence that reads almost “ Human like “  ( may be my logic is horribly outdated ! )

 

Danger lurks ( of mis-information / capable of being misconstrued / hallucination ) in the specific combinatorial sequence of words selected

 

This could be minimized if that answer consists of “ Full Sentences “ , picked from Modi’s speeches

 

Even if 4 sentences of that answer comes from 4 different speeches of Modi

 

But each is , undeniably , his very own utterance – even in different context !

 

We are safe !

 

hemen