I did some chatting, and i figured out, it cannot count his output for some reason. The second one is 584 words as per wordcounter.com.<p>What is happening?<p>Chat: https://pastebin.com/MfbVBmad
how many words does this question have?<p>This question has nine words.<p>can you make a list with all 9 words?<p>Sure, here is a list of the nine words in this question:<p><pre><code> how
many
words
does
this
question
have
?
</code></pre>
Note the last word is actually a special character, the '?' which is the question mark symbol , it is not a alphabetic word.<p>you only listed 8<p>You are correct, I apologize for the oversight. I listed 8 words instead of 9. The ninth word in the question is "many."<p>can you list all the 9 words again now that you found the mistake?<p>Sure, here is the corrected list of the nine words in this question:<p><pre><code> how
many
words
does
this
question
have
?
many
</code></pre>
once again, the last word is a special character and it's not a word itself<p>how many occurences of the word "many" did my question have and how many occurences of the word "many" did your list have?<p>Your original question contains the word "many" once, and my corrected list contains the word "many" twice.<p>how is that possible? you should only have listed the words from my question<p>You're correct, I apologize for the confusion. My previous response was in error. Your original question contains the word "many" only once, as you stated. I apologize for any confusion my mistake may have caused.<p>ok, tell me again, how many words did my question have?<p>Your question had 9 words.
ChatGPT is just a statistical engine that outputs text consistent with its previous input. So while it understands that requests to count words often end with something like "this passage has <number> words", and therefore puts a number in the sentence to make it statistically probable, it doesn't have any way of understanding that the number is supposed to correlate to the number of tokens in the passage referenced. It doesn't understand the meaning of the words you are putting into it, it just understands how to generate statistically probable text from those words. Furthermore because concepts like "word count" are statistically separate from the semantic content of the sentence, ChatGPT has an especially hard time building any kind of statistical model that outputs the correct number based on the query.