Re:Large Language Models Are More Persuasive Than Incentiviz...
by koakuma-chan on Saturday May 17, 2025 @05:40PM
(#44017184)
I asked an LLM and it said "A stack is a data structure that follows the Last In, First Out (LIFO) principle. This means that the last element added to the stack is the first element to be removed."
[ Parent
| on HN ]
Re:Large Language Models Are More Persuasive Than Incentiviz...
by abtinf on Saturday May 17, 2025 @06:25PM
(#44017491)
It’s subtle but I would regard this as an incorrect answer.
The structure of the LLM answer is:
A is B; B exhibits property C.
The correct answer is:
A exhibits property C; B is the class of things with property C; therefore A is B.
There is a crucial difference between these two.
Re:Large Language Models Are More Persuasive Than Incentiviz...
by literalAardvark on Saturday May 17, 2025 @06:38PM
(#44017550)
This doesn't apply to all prompts, and the prompt was not provided. Natural language is a fickle thing.
Re:Large Language Models Are More Persuasive Than Incentiviz...
by moffkalast on Saturday May 17, 2025 @06:49PM
(#44017599)
This kind of pointless hair splitting is why people would rather talk to an LLM.
Re:Large Language Models Are More Persuasive Than Incentiviz...
by Benjammer on Saturday May 17, 2025 @10:37PM
(#44018565)
This kind of “hair splitting” is the foundation on current prompt engineering though…
Re:Large Language Models Are More Persuasive Than Incentiviz...
by Matthyze on Sunday May 18, 2025 @06:09AM
(#44020260)
I think you've read too much early Wittgenstein. That is simply not how people communicate.
Re:Large Language Models Are More Persuasive Than Incentiviz...
by hansmayer on Saturday May 17, 2025 @05:44PM
(#44017217)
Yikes:( I am so worried about the damage that will be caused by the misuse of these tools. Already a lot of young folks will just mindlessly trust whatever the magic oracle spits out at them. We need to go back to testing people with pen and paper I suppose.
Re:Large Language Models Are More Persuasive Than Incentiviz...
by Karrot_Kream on Saturday May 17, 2025 @06:00PM
(#44017320)
I read this and I see a common thinking fallacy, when someone is inclined to believe something a priori they fit the evidence to their a priori beliefs.
Re:Large Language Models Are More Persuasive Than Incentiviz...
by hansmayer on Saturday May 17, 2025 @06:18PM
(#44017444)
No, its fairly simple - I misread
Re:Large Language Models Are More Persuasive Than Incentiviz...
by jstanley on Saturday May 17, 2025 @05:47PM
(#44017234)
Why is that a bad answer?
Re:Large Language Models Are More Persuasive Than Incentiviz...
by hansmayer on Saturday May 17, 2025 @05:54PM
(#44017276)
Sorry - I misread the LLM answer - actually the LLM produced a correct answer here
Re:Large Language Models Are More Persuasive Than Incentiviz...
by louthy on Saturday May 17, 2025 @06:15PM
(#44017432)
> No it is not…
That’s a queue, not a stack. The LLM response was correct.
Re:Large Language Models Are More Persuasive Than Incentiviz...
by danielbln on Saturday May 17, 2025 @06:16PM
(#44017439)
But a stack is commonly LIFO, not FIFO?!
Re:Large Language Models Are More Persuasive Than Incentiviz...
by idonotknowwhy on Saturday May 17, 2025 @07:07PM
(#44017674)
This reads like a line from a QwQ or Qwen3 CoT chain :)
Re:Large Language Models Are More Persuasive Than Incentiviz...
by koakuma-chan on Saturday May 17, 2025 @05:46PM
(#44017232)
I mean, is it wrong? It seems correct. Unless I'm missing something.
Re:Large Language Models Are More Persuasive Than Incentiviz...
by hansmayer on Saturday May 17, 2025 @05:56PM
(#44017293)
Oops, my bad. I seem to have misread. Sorry.
Re:Large Language Models Are More Persuasive Than Incentiviz...
by thinkcritical on Saturday May 17, 2025 @06:01PM
(#44017326)
No, a stack is LIFO like it said. A queue is FIFO or in other words LILO “Last In Last Out”.
Re:Large Language Models Are More Persuasive Than Incentiviz...
by koakuma-chan on Saturday May 17, 2025 @06:32PM
(#44017521)
My last job was at the office. I had my work queue implemented as a stack of files. I would sit at my desk and, in an infinite loop, pop files from my stack and process them. Occasionally, my supervisor would come and push a new file onto my stack. A naive worker would think that, once I was done with my stack, I could finally get some sleep, but no. Our office implemented something called "work stealing," where, once I was done with my own work, I had to visit a random co-worker and pop files from their stack.