It is slightly confusing to me why everyone was insisting ChatGPT was a threat to Google and that it would replace Google search. They don’t seem like comparable products.
If you imagine users asking questions to Google just to recieve a bunch of crappy listicles or the wiki page, versus ChatGPT, it makes more sense.
ChatGPT enables you to have a dialogue to ask follow up questions, more detail or summarize information in two sentences. Google can’t compete with that using a page-rank algorithm alone. It is incredibly powerful and it’s getting exponentially better.
I’d caution anyone who just dismisses it by calling it a chatbot or says it hallucinates too much. I found the accuracy between 3.5 to 4 pretty astonishing to the point where I now fear the AGI apocalypse.
Honestly for certain things that I was already knowledgable about, it provided a really great and accurate summary when I asked. Other times not so much, so I don’t feel comfortable using it for research like that.
I think if it could base its output on real sources and direct you to them it would be a bit better
It’s summarizing it’s own google-esque searches. Trusting it is like trusting a book report by a kid that pulled out some sentences. Trusting it is like trusting the Google result excerpts without acknowledging it can miss the nuance of the writer and report the exact opposite answer. These do not lie, they are just less than accurate summaries.
CHATGPT, although it sometimes lie so need to take it with a grain of salt
chatgpt will make up believable bs on spot if you ask it about anything remotely obscure
It is slightly confusing to me why everyone was insisting ChatGPT was a threat to Google and that it would replace Google search. They don’t seem like comparable products.
If you imagine users asking questions to Google just to recieve a bunch of crappy listicles or the wiki page, versus ChatGPT, it makes more sense.
ChatGPT enables you to have a dialogue to ask follow up questions, more detail or summarize information in two sentences. Google can’t compete with that using a page-rank algorithm alone. It is incredibly powerful and it’s getting exponentially better.
I’d caution anyone who just dismisses it by calling it a chatbot or says it hallucinates too much. I found the accuracy between 3.5 to 4 pretty astonishing to the point where I now fear the AGI apocalypse.
Honestly for certain things that I was already knowledgable about, it provided a really great and accurate summary when I asked. Other times not so much, so I don’t feel comfortable using it for research like that.
I think if it could base its output on real sources and direct you to them it would be a bit better
It can help, but it’s data is sometimes too old.
LLMs are far better at writing things that look like answers than they are at writing actual answers.
All ChatGPT results are made up, some of them just happen to be correct guesses.
Hah. Imagine if you went to Wikipedia and had to account for that 40% of all info on there was straight made up. Like the Scots wikipedia.
“lie”
It’s summarizing it’s own google-esque searches. Trusting it is like trusting a book report by a kid that pulled out some sentences. Trusting it is like trusting the Google result excerpts without acknowledging it can miss the nuance of the writer and report the exact opposite answer. These do not lie, they are just less than accurate summaries.