3 ms·

Those are sources, but there's still no evidence that the sources are real (the first one seems to be, from a quick google, but still), that the source actually

by ysavir 4y ago

Those are sources, but there's still no evidence that the sources are real (the first one seems to be, from a quick google, but still), that the source actually supports what ChatGPT is saying, or that ChatGPT understood the source correctly. Using ChaptGPT for research is interesting, and I think it's a cool way to learn new things. But I think if someone is going to share that information with someone else, it's appropriate to manually find and cite creditable sources, at least sufficiently enough that the ChatGPT aspect becomes irrelevant in a way other than discoverability.

nerpderp82 4y ago
I use ChatGPT to quickly learn a new subject, and it hallucinates about half the sources it cites. It will often make up a paper title that is a mashup of two or more papers. It usually gets the researchers in a field correct, so it does provide enough for one to start doing their own research and it does usually point one to the good survey papers on a topic. At least the https://www.usanpn.org/usa-national-phenology-network https://www.usanpn.org/usa-national-phenology-network exists!
dsfyu404ed 4y ago
>but there's still no evidence that the sources are real Sounds like it's successfully approximating a human (I wish I was joking). The current "state of the art" for human commenters who are acting in bad faith is to google their position and cite articles and papers that don't even support their claims knowing full well that people will have to waste time reading your BS sources to disprove you and that even if they do you've still one because you've got them to stop arguing against your position and instead argue against your sources. These miscreants are often successful in using their false appeals to authority to convert people to their side. Of course ChatGPT doesn't know what "bad faith" is but regardless it's almost performing on their level. We are marching toward a world where dishonest spinsters and BSers pretending to be experts are going to be replaced with bots.
- slowmovintarget 4y ago
  A massive increase in noise to a fragile information propagation system society has come to rely on (mostly for ill, it seems). What could go wrong? Add that to the context of so many existential issues we need clear communication on (freedoms, the global economy, reformation of society, climate change, Putin and Jinping's last gasps for "legacy"). All the while dodging machine generated hallucinations of off-by-that-much pronouncements on the world around us. We have a perfect storm brewing.
- dragonwriter 4y ago
  > Of course ChatGPT doesn't know what "bad faith" is but regardless it's almost performing on their level. Its trained largely on the public internet. That it reproduces bad faith argument naturally and fluently should not be surprising.

by ysavir 4y ago

nerpderp82 4y ago

I use ChatGPT to quickly learn a new subject, and it hallucinates about half the sources it cites. It will often make up a paper title that is a mashup of two or more papers. It usually gets the researchers in a field correct, so it does provide enough for one to start doing their own research and it does usually point one to the good survey papers on a topic. At least the https://www.usanpn.org/usa-national-phenology-network https://www.usanpn.org/usa-national-phenology-network exists!

dsfyu404ed 4y ago

>but there's still no evidence that the sources are real Sounds like it's successfully approximating a human (I wish I was joking). The current "state of the art" for human commenters who are acting in bad faith is to google their position and cite articles and papers that don't even support their claims knowing full well that people will have to waste time reading your BS sources to disprove you and that even if they do you've still one because you've got them to stop arguing against your position and instead argue against your sources. These miscreants are often successful in using their false appeals to authority to convert people to their side. Of course ChatGPT doesn't know what "bad faith" is but regardless it's almost performing on their level. We are marching toward a world where dishonest spinsters and BSers pretending to be experts are going to be replaced with bots.

slowmovintarget 4y ago
A massive increase in noise to a fragile information propagation system society has come to rely on (mostly for ill, it seems). What could go wrong? Add that to the context of so many existential issues we need clear communication on (freedoms, the global economy, reformation of society, climate change, Putin and Jinping's last gasps for "legacy"). All the while dodging machine generated hallucinations of off-by-that-much pronouncements on the world around us. We have a perfect storm brewing.
dragonwriter 4y ago
> Of course ChatGPT doesn't know what "bad faith" is but regardless it's almost performing on their level. Its trained largely on the public internet. That it reproduces bad faith argument naturally and fluently should not be surprising.