TL;DR
Prime made for students and young adults
- Fast, free delivery for dorm and study essentials
- Prime Video and Amazon Music included
- Member-only deals
Astral Codex Ten published an October 2026 links roundup spanning AI-generated art and poetry, marriage statistics, and a bot’s win in a forecasting competition. The author cautions that the links have not all been independently verified and says some claims discussed in the post may be wrong or misleading.
Astral Codex Ten has published its October 2026 links roundup, collecting items on AI-generated art and literary translation, a bot’s win in a forecasting competition, and disputes over claims about marriage. The post is a commentary roundup rather than an independently reported investigation: its author says they have not verified every link and warns that some may be wrong or misleading.
Among the items highlighted is a report that an AI bot called Laertes won the Metaculus Cup forecasting competition in September. The roundup cites an Economist account saying Laertes was developed by Jeffrey Liang, who reportedly spent fewer than 150 hours and a few thousand dollars on computing power and data. The post also names other high-performing bots and says one of them, FutureSearch, is publicly available. These details are presented through the cited reporting, not as results independently checked in the roundup.
The author also discusses experiments in AI-assisted translation and art. One example is a rendering of the opening of the Chinese novel Romance of the Three Kingdoms into rhyming English pentameter; the author says they do not speak Chinese and therefore cannot judge its accuracy. Another example reworks imagery associated with the painter Hieronymus Bosch. The post itself raises a qualification: the work draws heavily on a human artist, which complicates whether it should be treated as AI art or as an instance of AI enabling human creative work.
A separate entry challenges a post arguing that marriage is a bad idea for most people. The roundup says that post characterized only 18% of marriages as non-miserable, based on a survey that classified every answer except “very happy” as miserable. The author argues that this definition groups “pretty happy” marriages with unhappy ones and says their reading of the General Social Survey indicates 61% of respondents described their marriage as “very happy.” They report checking the survey figure but say they have not confirmed other alleged errors in the argument.
AI’s Forecasting Win Gets Attention
The roundup brings together developments that readers may otherwise encounter separately, but its most concrete news item is the reported Metaculus Cup result: a bot took first place in the competition for the first time, according to the source’s account. If the result and reported development details hold up, the entry adds a real-world example to ongoing discussion about how well AI systems can make forecasts compared with people.
The post does not establish that AI has reached general forecasting parity with humans. Its author notes a prior prediction about parity by the end of 2026, then points to the competition as a relevant development. One contest result alone does not settle that broader question. The roundup’s treatment of marriage statistics also illustrates why definitions and source data matter: a survey’s response categories can produce very different impressions depending on how an author groups them.
Popular right now
Ninja DualZone Air Fryer 10 Qt Smart 6-in-1 Grey (DZ950)
Ninja CREAMi Scoop & Swirl 13-in-1 Ice Cream and Soft Serve Maker, NC701
Roborock F25 GT Set Cordless Wet Dry Vacuum Cleaner and Mop with 2 Brushes
As an affiliate, we earn on qualifying purchases.
A Monthly Mix of Links
The post follows the format of a monthly links roundup: it shares items the author found notable, offers brief reactions, and sometimes disputes how a linked article presents its evidence. The entries cover a wide range, including Jeremy Bentham’s cookbook, an explanation for a strange phrase in a chemistry paper, commentary on American suburbs, and an account of the origins of modern polyamory.
Several entries concern how AI systems behave or might be used. Alongside creative experiments, the author recounts a claim that an older version of Claude could continue text made to resemble an internal company email. The post says the possibility that this reveals actual internal emails is probably false and cautions that this is not how training works. It treats the outputs instead as potentially evocative generated text, not proof of a model’s private views or access to confidential material.
The roundup also points to a report that AI forecasting startup Mantic raised $25 million in seed funding. It does not supply further detail about the funding terms or the company’s plans. Taken together, the entries are best read as pointers and commentary, with the author’s own caveats attached, rather than as a single verified account of October developments.
“I haven’t independently verified each link.”
— Astral Codex Ten author
Which Claims Still Need Checking
The author says commenters typically identify two or three wrong or misleading links in each roundup, on average, and that corrections are made as they are noticed. That figure describes the author’s experience with these posts; the source gives no detailed method or comparison basis. The author says they cannot guarantee all errors will have been caught by the time a reader sees the post.
Several specific matters remain unverified in the material provided. The author says they checked the marriage survey’s “not too happy” figure but have not confirmed other possible errors identified with help from Claude. They also cannot assess the accuracy of the Chinese-to-English verse example, say the AI artwork’s dependence on human work complicates its classification, and caution against treating generated text about AI welfare as evidence of a model’s actual desires. The full competition methodology and the circumstances behind the Mantic funding announcement are not detailed in the roundup.
Corrections and Follow-Up Reporting
The author says they will correct errors as they see them and highlight important corrections later. Readers seeking to verify a particular claim will need to follow the original linked reporting, check the underlying sources where available, and look for updates to the roundup. No further results or correction schedule are specified in the post.
For the forecasting story, more information about competition performance and the methods used by the participating bots would help put the win in perspective. For the marriage dispute, the key next step is to examine the survey wording and underlying figures alongside the interpretation challenged by the roundup. The post does not say when additional details on either topic will be published.
Key Questions
What is the October 2026 links post?
It is an Astral Codex Ten monthly roundup that links to a range of articles and examples, with short comments from the author. It is not presented as a fully verified news report.
What AI development does the roundup highlight?
The post reports that Laertes took first place in the Metaculus Cup forecasting competition in September, citing an Economist account. The roundup does not independently document the competition results or establish broader AI-human forecasting parity.
What does the roundup dispute about marriage statistics?
It challenges a post that described 18% of marriages as non-miserable after treating every survey response short of “very happy” as miserable. The roundup says its author’s reading of the General Social Survey found that 61% described their marriage as “very happy”, while noting that other alleged errors remain unchecked.
Are all the links and claims verified?
No. The author explicitly says they have not independently verified every link, that errors may be identified after publication, and that they cannot guarantee all have been corrected when readers encounter the post.
Source: astralcodexten.com
