4 ms·

I'm kind of hopeful AI some wikipedia-inspired AI "oracle" will happen in the future. Why not use some central AI to separate fact from fiction (faster than tro

by crnkofe 2mo ago

I'm kind of hopeful AI some wikipedia-inspired AI "oracle" will happen in the future. Why not use some central AI to separate fact from fiction (faster than trolls and griefers can come to muddy the debate).

n4r9 2mo ago
Possibly, but is an LLM is capable of this?
- Lwerewolf 2mo ago
  I mean, a big RAG setup with a smart LLM ought to do it, right? Or even just locally provided knowledge database - given the size of LLMs, what's a clone of wikipedia and whatever else you'd need?
- datakan 2mo ago
  No
- crnkofe 2mo ago
  Imo it can support the separation in meaningful ways. There are many political topics where opinions are flooding online forums much quicker than anyone with informed opinion can answer or digest them Having a quick reference for "verifying" statements would be useful. Sure LLMs are probabilistic but we don't have any issues using them for building software which runs as part of society infrastructure nowadays.
  - n4r9 2mo ago
    True, but software is very different to general knowledge. For example, it's easier to obtain reference documentation from the person who wrote the language. It's much easier to run tests to validate a given attempt. And the general discourse about how to implement a given solution or architecture is leagues apart from, say, the general discourse around the Israel-Palestine conflict.
SkyBelow 2mo ago
This is based on the assumption of facts existing. There are many studies, but each can be wrong and they can collectively show a bias. Even things of which are the most non-political of facts can have very strong biases. Look at the Millikan measurement of the electron and how in created confirmation bias and an anchoring effect on some property that has absolutely no real world significance to the things people are tribalistic about (aka, no political relevance). Now imagine the same applied to fields like economics or psychology which do have massive legal/political implications. For a different example, ask the question if X committed crime Y. There are cases where they weren't found guilty but it is reasonable to assume they did. But being found guilty doesn't make it a fact either, as some people are wrongly convicted. Some eventually are overturned, but even if it isn't, it still isn't a fact they committed a crime. Then there is the simple ambiguity of statements. Language generally can't support facts. It is why legalize, and programming code, and math's are effectively their own languages. For a simple example, consider the Betrand paradox(1). >Consider an equilateral triangle that is inscribed in a circle. Suppose a chord of the circle is chosen at random. What is the probability that the chord is longer than a side of the triangle? Is the answer 1/2, 1/3, or 1/4? Well, it is all three at once, depending upon what you meant by random. Now, imagine how this impacts things like research studies, where the randomness is much harder to quantify and there is constant pressure to p hack a result. 1. https://en.wikipedia.org/wiki/Bertrand_paradox_(probability) https://en.wikipedia.org/wiki/Bertrand_paradox_(probability)
- pjc50 2mo ago
  > Look at the Millikan measurement of the electron and how in created confirmation bias and an anchoring effect on some property Could you elaborate on this please? But yes, the underlying argument is sound: extracting facts from the messy phenomenological reports about the world is extremely difficult, and cannot be done by simply taking a brain in a jar and having it think extremely hard.
  - SkyBelow 2mo ago
    Incorrect calculation led to some wrong value, I think electric charge. Subsequent findings showed a general trend towards a correct value. The general trend indicates that findings that were too far off were scared to be published for being wrong and contradicting existing studies, thus only small refinements were published, causing the slow drift. Thus, the first well accepted study anchors a value and it takes much more work to get a new value established as correct after the anchor is set. Especially in a case where the experiment design itself is still solid and it was more with specific experiment (measurements slightly off, other values not quite correct).
thinkcontext 2mo ago
Isn't this what Elon Musk has unironically claimed he is doing with grokipedia?
- ceejayoz 2mo ago
  https://www.lawfaremedia.org/article/grokipedia-stopped-reviewing-edits-in-april.-it-didn-t-tell-anyone https://www.lawfaremedia.org/article/grokipedia-stopped-revi... > Grokipedia appears not to have updated any articles since April 24
- insane_dreamer 2mo ago
  yes, but at least you know what Grokipedia is, just like you know what you're going to get if you read Truth Social, or the Jacobin. but because the content is assimilated by GPT/Claude/Gemini and then mixed in with everything else, the bias is much much difficult to spot, making it much more effective and reaching a much wider audience (i.e., those who never visit Grokipedia)
inigyou 1mo ago
That's called chatgpt. You can influence it.

by crnkofe 2mo ago

n4r9 2mo ago

Possibly, but is an LLM is capable of this?

Lwerewolf 2mo ago
I mean, a big RAG setup with a smart LLM ought to do it, right? Or even just locally provided knowledge database - given the size of LLMs, what's a clone of wikipedia and whatever else you'd need?
datakan 2mo ago
No
crnkofe 2mo ago
Imo it can support the separation in meaningful ways. There are many political topics where opinions are flooding online forums much quicker than anyone with informed opinion can answer or digest them Having a quick reference for "verifying" statements would be useful. Sure LLMs are probabilistic but we don't have any issues using them for building software which runs as part of society infrastructure nowadays.
- n4r9 2mo ago
  True, but software is very different to general knowledge. For example, it's easier to obtain reference documentation from the person who wrote the language. It's much easier to run tests to validate a given attempt. And the general discourse about how to implement a given solution or architecture is leagues apart from, say, the general discourse around the Israel-Palestine conflict.

SkyBelow 2mo ago

This is based on the assumption of facts existing. There are many studies, but each can be wrong and they can collectively show a bias. Even things of which are the most non-political of facts can have very strong biases. Look at the Millikan measurement of the electron and how in created confirmation bias and an anchoring effect on some property that has absolutely no real world significance to the things people are tribalistic about (aka, no political relevance). Now imagine the same applied to fields like economics or psychology which do have massive legal/political implications. For a different example, ask the question if X committed crime Y. There are cases where they weren't found guilty but it is reasonable to assume they did. But being found guilty doesn't make it a fact either, as some people are wrongly convicted. Some eventually are overturned, but even if it isn't, it still isn't a fact they committed a crime. Then there is the simple ambiguity of statements. Language generally can't support facts. It is why legalize, and programming code, and math's are effectively their own languages. For a simple example, consider the Betrand paradox(1). >Consider an equilateral triangle that is inscribed in a circle. Suppose a chord of the circle is chosen at random. What is the probability that the chord is longer than a side of the triangle? Is the answer 1/2, 1/3, or 1/4? Well, it is all three at once, depending upon what you meant by random. Now, imagine how this impacts things like research studies, where the randomness is much harder to quantify and there is constant pressure to p hack a result. 1. https://en.wikipedia.org/wiki/Bertrand_paradox_(probability) https://en.wikipedia.org/wiki/Bertrand_paradox_(probability)

pjc50 2mo ago
> Look at the Millikan measurement of the electron and how in created confirmation bias and an anchoring effect on some property Could you elaborate on this please? But yes, the underlying argument is sound: extracting facts from the messy phenomenological reports about the world is extremely difficult, and cannot be done by simply taking a brain in a jar and having it think extremely hard.
- SkyBelow 2mo ago
  Incorrect calculation led to some wrong value, I think electric charge. Subsequent findings showed a general trend towards a correct value. The general trend indicates that findings that were too far off were scared to be published for being wrong and contradicting existing studies, thus only small refinements were published, causing the slow drift. Thus, the first well accepted study anchors a value and it takes much more work to get a new value established as correct after the anchor is set. Especially in a case where the experiment design itself is still solid and it was more with specific experiment (measurements slightly off, other values not quite correct).

thinkcontext 2mo ago

Isn't this what Elon Musk has unironically claimed he is doing with grokipedia?

ceejayoz 2mo ago
https://www.lawfaremedia.org/article/grokipedia-stopped-reviewing-edits-in-april.-it-didn-t-tell-anyone https://www.lawfaremedia.org/article/grokipedia-stopped-revi... > Grokipedia appears not to have updated any articles since April 24
insane_dreamer 2mo ago
yes, but at least you know what Grokipedia is, just like you know what you're going to get if you read Truth Social, or the Jacobin. but because the content is assimilated by GPT/Claude/Gemini and then mixed in with everything else, the bias is much much difficult to spot, making it much more effective and reaching a much wider audience (i.e., those who never visit Grokipedia)

inigyou 1mo ago

That's called chatgpt. You can influence it.