
| News | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
I started trying to get LLMs to do math in July 2020, through the game "AI Dungeon," one of the earliest applications powered by GPT-3. I first got GPT-3 to produce a correct proof (of Fermat's Little Theorem) in April 2022. At the time I did not think they would become useful for math research in the near term. This changed when the first reasoning models were released: on February 1, 2025, I wrote that the model o3-mini-high “clearly has passed the threshold of genuine usefulness” for research, while still making many, many mistakes. Since then, the models have improved, and ChatGPT 5.2 Pro (released in December 2025) can regularly provide reasonable proofs of lemmas that I would characterize as “involved but routine for experts,” though it still makes many errors. And I have been using Codex, OpenAI's coding/computer use agent, for scientific computing tasks I would not have considered attempting a few months ago.
...
I think I have been underrating the pace of model improvements. In March 2025 I made a bet with Tamay Besiroglu, cofounder of RL environment company Mechanize, that AI tools would not be able to autonomously produce papers I judge to be at a level comparable to that of the best few papers published in 2025, at comparable cost to human experts, by 2030. I gave him 3:1 odds at the time; I now expect to lose this bet. | |
Submitted at Today, 02:08 AM by lurk on my face | |
1 Comment | |
The state rescinded its request to dismiss a sexual abuse lawsuit after a judge became aware of New York Focus’s findings. | |
Submitted at Today, 12:50 AM by sleeppoor | |
An internal Department of Homeland Security document shows how ICE plans to cram thousands of detained human beings inside a Georgia warehouse. | |
Submitted at Today, 12:47 AM by sleeppoor | |
The operation set off a wave of violence, with torched cars and gunmen blocking highways in more than half a dozen states. | |
Submitted at Today, 12:17 AM by sleeppoor | |
Submitted at Yesterday, 04:50 PM by Grief Bacon | |
As smoking rates fall in the U.S., startups and influencers are pushing the purported cognitive and health benefits of indulging in nicotine. | |
Submitted at Yesterday, 07:48 PM by sleeppoor | |
Civil rights groups and parents gathered in front of the Quakertown Police Department Saturday demanding answers for the violent confrontation Friday. | |
Submitted at Yesterday, 11:00 AM by sleeppoor | |
Federal prosecutors have arraigned four people in New Jersey, with a fifth at large in Colombia | |
Submitted at Yesterday, 01:54 AM by B. Weed | |
A mother of three in Utah self-published a children’s book to help her sons cope with the death of their father. Now she is on trial accused of murder | |
Submitted at 02-21-2026, 06:51 PM by sleeppoor | |
String of embarrassing defeats for prosecutors as experts condemn DoJ effort to cast people as ‘violent perpetrators’ | |
Submitted at 02-21-2026, 06:01 PM by B. Weed | |
Incarceration has become a central issue in Western Arkansas, where the governor wants to build a new prison that has angered and mobilized residents. | |
Submitted at 02-21-2026, 07:59 AM by sleeppoor | |
Carr wants broadcasters to run patriotic specials and PSAs. | |
Submitted at 02-20-2026, 10:17 PM by sleeppoor | |
Supreme Court Justice Brett Kavanaugh, in his dissent, warned that "the refund process" for tariffs "is likely to be a 'mess,' " citing oral arguments | |
Submitted at 02-20-2026, 04:54 PM by sleeppoor | |
A federal judge has rejected Tesla’s bid to overturn a $243 million jury verdict over a fatal 2019 Autopilot crash... | |
Submitted at 02-20-2026, 04:51 PM by sleeppoor | |
Retired wildlife biologist Bob Bancroft describes the elimination of the division and today’s changes to the department as “absolutely disgusting.” | |
Submitted at 02-20-2026, 04:28 PM by sleeppoor | |
Publisher Finji says that TikTok has been using generative AI to modify its ads on the platform without permission and pushing those ads to its users without Finji's knowledge, including at least one ad that was modified to include a racist, sexualized stereotype of one of Finji's characters. | |
Submitted at 02-20-2026, 04:08 PM by sleeppoor | |
Butler Mayor Wesley R. Dingus faces two misdemeanor voyeurism charges following a Richland County investigation. The case unfolds as he remains on bond in a separate felony matter. | |
Submitted at 02-20-2026, 04:25 AM by sleeppoor | |
The unusual looking fossil is estimated to be a few hundred million years old dating to the Carboniferous period. | |
Submitted at 02-19-2026, 04:37 PM by sleeppoor | |
The Seoul Central District Court on Thursday sentenced former President Yoon Suk Yeol to life imprisonment, finding him guilty of leading an insurrection. "It i | |
Submitted at 02-19-2026, 04:34 PM by sleeppoor | |
Andrew Mountbatten-Windsor has been arrested on suspicion of misconduct in public office. The former Prince Andrew was stripped of his royal titles because of his links to convicted sex offender Jeffrey Epstein. | |
Submitted at 02-19-2026, 03:08 PM by Grief Bacon | |

I started trying to get LLMs to do math in July 2020, through the game "AI Dungeon," one of the earliest applications powered by GPT-3. I first got GPT-3 to produce a correct proof (of Fermat's Little Theorem) in April 2022. At the time I did not think they would become useful for math research in the near term. This changed when the first reasoning models were released: on February 1, 2025, I wrote that the model o3-mini-high “clearly has passed the threshold of genuine usefulness” for research, while still making many, many mistakes. Since then, the models have improved, and ChatGPT 5.2 Pro (released in December 2025) can regularly provide reasonable proofs of lemmas that I would characterize as “involved but routine for experts,” though it still makes many errors. And I have been using Codex, OpenAI's coding/computer use agent, for scientific computing tasks I would not have considered attempting a few months ago.
...
I think I have been underrating the pace of model improvements. In March 2025 I made a bet with Tamay Besiroglu, cofounder of RL environment company Mechanize, that AI tools would not be able to autonomously produce papers I judge to be at a level comparable to that of the best few papers published in 2025, at comparable cost to human experts, by 2030. I gave him 3:1 odds at the time; I now expect to lose this bet.
The state rescinded its request to dismiss a sexual abuse lawsuit after a judge became aware of New York Focus’s findings.
An internal Department of Homeland Security document shows how ICE plans to cram thousands of detained human beings inside a Georgia warehouse.
The operation set off a wave of violence, with torched cars and gunmen blocking highways in more than half a dozen states.
As smoking rates fall in the U.S., startups and influencers are pushing the purported cognitive and health benefits of indulging in nicotine.
Civil rights groups and parents gathered in front of the Quakertown Police Department Saturday demanding answers for the violent confrontation Friday.
Federal prosecutors have arraigned four people in New Jersey, with a fifth at large in Colombia
A mother of three in Utah self-published a children’s book to help her sons cope with the death of their father. Now she is on trial accused of murder
String of embarrassing defeats for prosecutors as experts condemn DoJ effort to cast people as ‘violent perpetrators’
Incarceration has become a central issue in Western Arkansas, where the governor wants to build a new prison that has angered and mobilized residents.
Carr wants broadcasters to run patriotic specials and PSAs.
Supreme Court Justice Brett Kavanaugh, in his dissent, warned that "the refund process" for tariffs "is likely to be a 'mess,' " citing oral arguments
A federal judge has rejected Tesla’s bid to overturn a $243 million jury verdict over a fatal 2019 Autopilot crash...
Retired wildlife biologist Bob Bancroft describes the elimination of the division and today’s changes to the department as “absolutely disgusting.”
Publisher Finji says that TikTok has been using generative AI to modify its ads on the platform without permission and pushing those ads to its users without Finji's knowledge, including at least one ad that was modified to include a racist, sexualized stereotype of one of Finji's characters.
Butler Mayor Wesley R. Dingus faces two misdemeanor voyeurism charges following a Richland County investigation. The case unfolds as he remains on bond in a separate felony matter.
The unusual looking fossil is estimated to be a few hundred million years old dating to the Carboniferous period.
The Seoul Central District Court on Thursday sentenced former President Yoon Suk Yeol to life imprisonment, finding him guilty of leading an insurrection. "It i
Andrew Mountbatten-Windsor has been arrested on suspicion of misconduct in public office. The former Prince Andrew was stripped of his royal titles because of his links to convicted sex offender Jeffrey Epstein.