Verbosity and the Exploding Brain

[Thales750]

I thought it might be fun to compare some of the things we find problematic, or at least annoying, in our interactions with the AIs.

As the title states, sometimes all I want is a simple yes or no, and I am delivered a dissertation. Other times, I question a bit of code and it writes a new procedure, or maybe even an entire module, plus an encyclopedia of explanations.

And I can't tell you how many of these conversation were initiated because the GPT, for no reason other than it embellished some code into a full chaotic symphony resembling a spiraling vortex.

Until all that is left is an undecipherable cacophony.

You look at this code, and your brain says why, why, why, for the love of God does this machine include so much fluff?
And then your brain explodes.

 

[The_Doc_Man]

As the title states, sometimes all I want is a simple yes or no, and I am delivered a dissertation.

It's a strategy to frustrate you and thereby dissuade you from asking any more annoying questions. We sesquipedalianists know this method well. Whom do you think taught it to the AI during their training sessions?

 

[AccessBlaster]

Verbosity and the Exploding Brain

I sympathize with AI, it's probably tired of your conspiracy theories. Just kidding

 

[Thales750]

I sympathize with AI, it's probably tired of your conspiracy theories. Just kidding

They aren't theories.

 

[Isaac]

s the title states, sometimes all I want is a simple yes or no, and I am delivered a dissertation

yes! sometimes it is overkill. I like it better when it gives me a shorter answer, then nudges me to ask the best follow up question - which it always does, but sometimes after 'too much' of an answer.

You look at this code, and your brain says why, why, why, for the love of God does this machine include so much fluff?
And then your brain explodes.

I'd rather it err on the side of too much rather than not enough, but I do sometimes wish for shorter answers - but thats mostly for non-code situations. for me usually with code it gives me roughly the right amount of response to my personal taste.

My biggest imperfection I see in chatgpt continues to be its wish to compliment me rather than challenge me. Recently it has been challenging me a little bit more than usual in some of my book sales and I'm not sure if it finally learned me well enough to know I want to be challenged or something changed on its programming?

Right now I'm noticing that after daily sales of $20+ in my Amazon KDP book sales, yesterday was $3 and today is nothing so far. I figure it's one of two things: either "people just aren't buying on christmas eve/day but it will go back to normal soon", or else "my entire momentum of the last 45-50 days was entirely ABOUT christmas and that's it - and the momentum is now over". Desperately hoping it's the former, of course. I think so, because all of November was strong and growing, and that's still quite a while prior to Christmas, so hopefully it had little to do with Christmas

 

[The_Doc_Man]

You look at this code, and your brain says why, why, why, for the love of God does this machine include so much fluff?

Because it trained on your articles?