> People can reasonably expect Wikipedia to have factual data even though it sometimes contains inaccuracies.
I just straight-up don't agree with this, nor with the idea that what people consider "facts" are nearly as reliable as is implied. What we actually refer to via "fact" is "consensus". Truth is an apriori concept whereas we're discussing posteriori claims. Any "reasonable" ai would give an indication of degree of certainty, and there's no reliable or consensus-driven methodology to produce this manually, let alone automatically. The closest we come is the institution of "science" which can't even—as it stands—reliably address the vast majority of claims made about the world today.
And this is even before discussing the thorny topic of the ways in which language binds to reality, to which I refer you to Wittgenstein, a person likely far more intelligent and epistemologically honest than anyone influencing AI work today.
Yes, wikipedia does tend to cohere with reality, or at least it sometimes does in my experience. That observation is wildly different from an expectation that it does in the present or will in the future reflect reality. Futhermore it's not terribly difficult to find instances where it's blatantly not correct. For instance, I've been in a wikipedia war over whether or not the Soviet Union killed 20 million christians for being christians (spoiler: they did not, and this is in fact more people than died in camps or gulags over the entire history of the state). However, because there are theologists at accredited universities that have published this claim, presumably with a beef against the soviet union for whatever reason (presumably "anticommunism"), it's considered within the bounds of accuracy by wikipedia.
EDIT0: I'm not trying to claim wikipedia isn't useful; I read it every day and generally take what it says to be meaningful and vaguely accurate. but the idea that you should trust what you read on it seems ridiculous. As always, it's only as reliable as the sources it cites, which are only as reliable as the people and institutions that produce that cited work.
EDIT1: nice to see someone else from western mass on here; cheers. I grew up in the berkshires.
For those not in the know, the Soviet Union was an officially atheist empire and explicitly anti religious, foremost anti christian. I don't know if the poster above me is denying this or if he/she considers it general knowledge which needn't be mentioned.
Treatment of christians and christianity varied widely over the lifetime of the soviet union. That the soviet union was "atheist" is an incorrect reduction of the situation (albeit one reinforced by the propaganda both in the soviet union and especially in the US)
In practice, starting with Stalin the party mostly let the church continue unmolested.
"The Great Patriotic War changed Joseph Stalin’s position on the Orthodox Church. In 1943, after Stalin met with loyal Metropolitans, the government let them choose a new Patriarch, with government support and funding, and permitted believers to celebrate Easter, Christmas and other holidays. Stalin legalized Orthodoxy once again."
"Unmolested" during the war and until Stalin's death. After that, Khrushchev closed churches and started the anti christian campaigning again. The mass murder of "state enemies" mostly ended with Stalin's death, but the church and Christians were still molested, even though they weren't tortured to death.
I just straight-up don't agree with this, nor with the idea that what people consider "facts" are nearly as reliable as is implied. What we actually refer to via "fact" is "consensus". Truth is an apriori concept whereas we're discussing posteriori claims. Any "reasonable" ai would give an indication of degree of certainty, and there's no reliable or consensus-driven methodology to produce this manually, let alone automatically. The closest we come is the institution of "science" which can't even—as it stands—reliably address the vast majority of claims made about the world today.
And this is even before discussing the thorny topic of the ways in which language binds to reality, to which I refer you to Wittgenstein, a person likely far more intelligent and epistemologically honest than anyone influencing AI work today.
Yes, wikipedia does tend to cohere with reality, or at least it sometimes does in my experience. That observation is wildly different from an expectation that it does in the present or will in the future reflect reality. Futhermore it's not terribly difficult to find instances where it's blatantly not correct. For instance, I've been in a wikipedia war over whether or not the Soviet Union killed 20 million christians for being christians (spoiler: they did not, and this is in fact more people than died in camps or gulags over the entire history of the state). However, because there are theologists at accredited universities that have published this claim, presumably with a beef against the soviet union for whatever reason (presumably "anticommunism"), it's considered within the bounds of accuracy by wikipedia.
EDIT0: I'm not trying to claim wikipedia isn't useful; I read it every day and generally take what it says to be meaningful and vaguely accurate. but the idea that you should trust what you read on it seems ridiculous. As always, it's only as reliable as the sources it cites, which are only as reliable as the people and institutions that produce that cited work.
EDIT1: nice to see someone else from western mass on here; cheers. I grew up in the berkshires.
EDIT2: to add on to the child comment, wikipedia is occasionally so hilariously unreliable it makes the news. Eg https://www.theguardian.com/uk-news/2020/aug/26/shock-an-aw-...