Loved the Boxing day trivia, and the thought of a Random Trivia section where you do the hard work of answering questions I have had forever, but been too lazy/busy/occupied to actually look up.
HA ! No, I am actually deeply distrustful of AI (specifically LLMs - e.g. ChatGPT, Claude, etc.) because it hallucinates with authority. So actually I think it is absolutely terrible at researching brand new items unless you have a way to check them against an authoritative source. ChatGPT will give you maybe 80% good things, but 20% you have no idea it hallucinates way off into the deep end.
But, if you go to an authoritative source, why use the LLM? :)
Actually it's hugely beneficial to build/finetune an authoritative datasource using raw unstructured data and LLMs. That is an excellent use of it...
Meanwhile - I will accept your answer to "What is the origin of Boxing Day" and will not check it, because there's a certain inherent trust I have that a human did the checking. It's actually weird, because a human is more prone to bias/making mistakes in general.
In this instance, I could get the same answer from ChatGPT, but I would spend the extra time to fact check it because I do not trust it not to hallucinate.
completely agree. I was actually also thinking about another (paid) future section/challenge regarding this, smth like "who's answer is better, mine or gpt?" :D
Loved the Boxing day trivia, and the thought of a Random Trivia section where you do the hard work of answering questions I have had forever, but been too lazy/busy/occupied to actually look up.
Thanks, I'm glad so many ppl liked that one!
And ooo big thanks for that suggestion! 😃 (but wouldn't u personally just "apply some AI" to get ur answer? 😜)
HA ! No, I am actually deeply distrustful of AI (specifically LLMs - e.g. ChatGPT, Claude, etc.) because it hallucinates with authority. So actually I think it is absolutely terrible at researching brand new items unless you have a way to check them against an authoritative source. ChatGPT will give you maybe 80% good things, but 20% you have no idea it hallucinates way off into the deep end.
But, if you go to an authoritative source, why use the LLM? :)
Actually it's hugely beneficial to build/finetune an authoritative datasource using raw unstructured data and LLMs. That is an excellent use of it...
Meanwhile - I will accept your answer to "What is the origin of Boxing Day" and will not check it, because there's a certain inherent trust I have that a human did the checking. It's actually weird, because a human is more prone to bias/making mistakes in general.
In this instance, I could get the same answer from ChatGPT, but I would spend the extra time to fact check it because I do not trust it not to hallucinate.
completely agree. I was actually also thinking about another (paid) future section/challenge regarding this, smth like "who's answer is better, mine or gpt?" :D
Beautifully written. Thanks for the boxing day trivia, we were so wrong about it.
Thanks so much Nayan - I'm glad you found it interesting 😄
Thank you for the shoutout!! I'm not sure which of the Substacks I follow led me to you, unforch!
haha no worries, all that matters is that it happened - enjoy ur hols! 😄