7 ms·
> I don’t trust this post The post appended the raw data provided by GPT, allowing you to verify the integrity of the data. This makes the post trustworthy fro
by inductive_magic 3y ago
> I don’t trust this post
The post appended the raw data provided by GPT, allowing you to verify the integrity of the data. This makes the post trustworthy from a methodological pov.
> I don’t trust anything published on the Internet after Llms went mainstream.
You always had to verify the integrity of the data and methods used in any publication, regardless of the medium. The responsibility of both authors and readers hasn't changed. If you took things for granted before LLMs, you shouldn't have, and if you don't trust trustworthy authors post LLMs, you should.
- danparsonson 3y agoOne could argue that whereas once it was necessary to verify the source, now it is necessary to verify not just the source but also the LLM derivation of it, (which may be subtly mangled) - and the source may no longer be readily apparent.
- xarope 3y agoI think this is a good thing. In the past you would remind people, hey, after you find your wiki article, "please" go verify. But wiki was "good" enough most of the time, that people found verification to be often time redundant. But now with LLMs everywhere, people will realize it is necessary to verify.
- meheleventyone 3y agoExcept that isn’t what will happen and is part of the reason we tried to have majority trustworthy sources.
- TheSpiceIsLife 3y agoYou're optimistic!
- xarope 3y agoPrepare for the worst, hope for the best!
- phanimahesh 3y agoI hope I can be this optimistic.
- krapp 3y agoHow do you verify when every source you could possibly refer to has also been poisoned by AI?
- rsynnott 3y agoOr just disregard any machine-generated stuff entirely, though of course it may not always be labeled as such.
- kortilla 3y ago> This makes the post trustworthy from a methodological pov. A post is not trustworthy if it’s reposting trash, even if it shows the source. > you took things for granted before LLMs, you shouldn't have, and if you don't trust trustworthy authors post LLMs, you should. The nature of how LLMs hallucinate is different from how garbage used to appear on the internet. Before LLMs there was a relatively good inverse correlation between quality and blatant bullshit. Not enough to pass the verification rigor required for an academic publication by any means, but enough that you didn’t have to second guess every single statement on every web page listing something as simple as book authors.
- criley2 3y agoHere's my radical take: humans hallucinate as much or more than the top LLMs. Heck, depending on the human, nearly everything they think and say is functionally a hallucination according to the metrics in use here. When it comes to what LLM's write, I find that LLM hallucinations are like self-driving car crashes. We are hyper-aware of the machine doing something that we ourselves do every single day and consider a normal defect of biological conscious.
- Scarblac 3y ago> Here's my radical take: humans hallucinate as much or more than the top LLMs. I can't believe that. LLMs always talk in the same confident tone, entirely regardless of what they're saying is true or not. What is true in the real world literally doesn't come into the equation. Whereas at least some of the time, humans will say that they're not sure and might be wrong, or otherwise sound less confident. And that's related to how true the thing they're saying is.
- pmarreck 3y ago> humans will say that they're not sure and might be wrong Is that so? https://innocenceproject.org/dna-exonerations-in-the-united-states/ https://innocenceproject.org/dna-exonerations-in-the-united-... These people were convicted by people who were 100% convinced their memory was correct. The DNA evidence, which is "harder" evidence, said otherwise, and in these cases, was exonerating. (There are hundreds, possibly thousands, of other cases like this by the way, where the imprisoned innocent is NOT yet exonerated, all based on overconfident eyewitness testimony that yet managed to convince a judge/jury.) There is also the well-known Dunning-Kruger effect, the cognitive bias where individuals with limited knowledge or expertise in a particular area tend to overestimate their competence and confidently assert their opinions. We've literally seen this countless times just since the 2016 US election, just watch literally any Jordan Klepper interview https://www.youtube.com/watch?v=LoZ2Lt_aCo8 https://www.youtube.com/watch?v=LoZ2Lt_aCo8 (honestly, this is a little too political for me to use as an example, but I ran out of time seeking out unbiased examples... Mandela Effect? Misplaced keys being common?) I'm afraid you're a little off, here, on your faith in humans not hallucinating memories and knowledge.
- otteromkram 3y agoI don't trust this comment.
- Aurornis 3y ago> If you took things for granted before LLMs, you shouldn't have, and if you don't trust trustworthy authors post LLMs, you should. Blaming LLMs for everything is becoming the preferred excuse for people who like to reject what they read and substitute their own beliefs instead. It’s true that LLMs hallucinate and are definitely not correct all the time, but the way people are using that as an opening to reject everything on the internet and elevate their own prior beliefs to the top is strange.
- yndoendo 3y agoMost people don't know humans hallucinate daily. Right now your brain is removing visual information by clipping out the image of your nose and replacing it with what the scene behind it theoretically looks like.
- ChatGTP 3y agoIMO ChatGPT 5 will just think the answer to all problems is going to be "humans hallucinate too?!?!". Maybe it will be right, but the best part? We'll continue to be reminded of this fact. Thanks for reminding us all.