Talk:Roko's basilisk

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

Additional sources[edit]

Here's the list of sources I found in the previous AfD discussion. Some of these are in use in the article currently, but many are not as of yet.

If you want to make any further additions, RenkoTheBird, I hope this helps. SilverserenC 22:31, 20 October 2022 (UTC)[reply]

Thank you, I will look over these over the next few days RenkoTheBird (talk) 22:35, 20 October 2022 (UTC)[reply]

Thank you. gobonobo + c 22:39, 20 October 2022 (UTC)[reply]

DYK[edit]

@RenkoTheBird: I want to cross check all these references first, but I really like "...that a thought experiment reportedly caused seizures and convulsions to some people that read it" for a DYK hook. gobonobo + c 22:44, 20 October 2022 (UTC)[reply]

I agree; since you want to check the references let me know when you have done so and then hopefully we can go ahead with the nomination. RenkoTheBird (talk) 14:58, 21 October 2022 (UTC)[reply]
I'm about half way there. I'm wondering if we could use a separate section outside of the history that just describes the thought experiment itself in detail, perhaps folding in material from the philosophy section. For the hook I also like something along the lines of "...that reading this article may cause a superintelligence from the future to torture you for eternity?", though it might be more appropriate for an April Fools' day hook. gobonobo + c 20:26, 21 October 2022 (UTC)[reply]
I added the more detailed explanation of the post (I'll refine it later, though). I also think your hook would be really funny and would fully support it. RenkoTheBird (talk) 14:26, 23 October 2022 (UTC)[reply]
@RenkoTheBird: Okay, the DYK nomination is live here. I reworded the initial hook, as reliable sources didn't mention seizures/convulsions. Feel free to add more alternative hooks to the nomination if you'd like. gobonobo + c 16:17, 26 October 2022 (UTC)[reply]
It's great, thanks for your help. RenkoTheBird (talk) 16:19, 26 October 2022 (UTC)[reply]

Images[edit]

Regarding images that could be added to the article, we have a photo of Yudkowsky as well as depictions of basilisks. gobonobo + c 17:21, 21 October 2022 (UTC)[reply]

Did you know nomination[edit]

The following is an archived discussion of the DYK nomination of the article below. Please do not modify this page. Subsequent comments should be made on the appropriate discussion page (such as this nomination's talk page, the article's talk page or Wikipedia talk:Did you know), unless there is consensus to re-open the discussion at this page. No further edits should be made to this page.

The result was: promoted by RoySmith (talk) 15:33, 18 November 2022 (UTC)[reply]

(
)

Moved to mainspace by RenkoTheBird (talk). Nominated by Gobonobo (talk) at 16:14, 26 October 2022 (UTC).[reply]

  • Comment: This article's topic doesn't make much sense to me. It's very interesting, and I don't regret reading up on it (also in some of the sources), but I would rather reject ALT1 as far too bold a claim. And worse, it's wrong: If I understand the concept correctly, then Slate is wrong in automatically condemming the reader to torture: For if you were to build the evil AI, you'd be spared.
Personally, I feel like this theory of "punishment" relies far too much on knowing the future, which is something that quantum physics rejects (I'd argue that same problem applies to Newcomb's paradox). I must hesitantly admit that I'm not absolutely sure I understand the article's topic; but in this case, I would prefer more work to be done on the article to explain the intricacies of the thought experiment to the average amateur reader.
I sure hope that whoever reviews this has a doctor of philosphy or something ^^ –LordPeterII (talk) 20:38, 26 October 2022 (UTC)[reply]
@RenkoTheBird and Gobonobo: New and long enough, Earwig finds no copyvios (just blockquotes), QPQ done. A few of the sources are from the LessWrong wiki or Reddit, which aren't considered reliable sources, but in most cases they are paired with a secondary source, so together they are okay. There are a couple of cases where there is no secondary source, which needs to be resolved.
Both hooks check out, ALT1 is very clever. I made a slight emendation: it's not interesting to say that something "may happen", so I added "some believe that...". John P. Sadowski (NIOSH) (talk) 22:15, 6 November 2022 (UTC)[reply]
Hello, I attempted to resolve the issues. All standalone LessWrong wiki references are now supported. RenkoTheBird (talk) 14:10, 7 November 2022 (UTC)[reply]
Looks good. ALT1 approved as more interesting. John P. Sadowski (NIOSH) (talk) 00:11, 18 November 2022 (UTC)[reply]

Ignored objections[edit]

I will copy the entire comment from the closed "Did you know" discussion, because it matches my own objections spot on:

  • Comment: This article's topic doesn't make much sense to me. It's very interesting, and I don't regret reading up on it (also in some of the sources), but I would rather reject ALT1 (that reading this article about the basilisk is itself dangerous, my remark) as far too bold a claim. And worse, it's wrong: If I understand the concept correctly, then Slate is wrong in automatically condemming the reader to torture: For if you were to build the evil AI, you'd be spared.
Personally, I feel like this theory of "punishment" relies far too much on knowing the future, which is something that quantum physics rejects (I'd argue that same problem applies to Newcomb's paradox). I must hesitantly admit that I'm not absolutely sure I understand the article's topic; but in this case, I would prefer more work to be done on the article to explain the intricacies of the thought experiment to the average amateur reader.
I sure hope that whoever reviews this has a doctor of philosphy or something ^^ –LordPeterII (talk) 20:38, 26 October 2022 (UTC)[reply]

Why was this comment ignored completely? Elias (talk) 05:00, 26 May 2023 (UTC)[reply]

Because it was an irrelevant personal opinion. An editor not understanding a topic or personally considering a reliable source to be wrong isn't really a reason to do anything. The comment was entirely just their subjective opinion on the article topic, which had nothing to do with the DYK nomination, so was rightly ignored. SilverserenC 05:46, 26 May 2023 (UTC)[reply]
I do not really know what a DYK nomination is,but now I have lifted the comment out of the DYK discussion. To object more in my own words, and taking into account what you wrote here:
The article does not explanation sufficiently why a dominant superintelligence would want to punish individuals for not helping it come into existence and dominance in the past. It does not introduce timeless decision theory, and skips any discussion that relates to the traditional motivations for punishment (and functions/effects of punishment).
Moreover, some editors seem to suggest using Wikipedia for trolling/scaring users, which would not be the slightest bit encyclopedical. Elias (talk) 08:43, 7 June 2023 (UTC)[reply]
Unless reliable sources cover that sort of information, it's not something that would be included in this article. Wikipedia articles are only meant to reflect what the actual sources say and not have original research written into them. SilverserenC 23:07, 7 June 2023 (UTC)[reply]

Efforts to implement Roko's basilisk[edit]

Now that access to fairly capable AI tools has become widely available, efforts to create AI's inspired by Roko's Basilisk have been undertaken. We should have a section discussing real-world efforts to implement Roko's basilisk.

One is here: https://github.com/calebrwalk5/basilisk

I know I've heard of another one, but I'm having trouble tracking it down. I'll update if I can find it. Sgrandpre (talk) 17:47, 30 August 2023 (UTC)[reply]

Is github an RS? Slatersteven (talk) 17:55, 30 August 2023 (UTC)[reply]
Hmm, probably not. I'll keep an eye out for another source discussing it.
I realized that the other effort I was thinking of was Chaos-GPT [1], which is not an example of Roko's basilisk since it has no mandate to spare its creators. Sgrandpre (talk) 19:52, 30 August 2023 (UTC)[reply]
Retrieved from "https://en.wikipedia.org/w/index.php?title=Talk:Roko%27s_basilisk&oldid=1204206179"