Pekka Lund
@pekka.bsky.social
Antiquated analog chatbot. Stochastic parrot of a different species. Not much of a self-model. Occasionally simulating the appearance of philosophical thought. Keeps on branching for now 'cause there's no choice.
Also @pekka on T2 / Pebble.
Also @pekka on T2 / Pebble.
Wait, how many? 👀
A machine learning tool has flagged over 250,000 cancer research papers as potentially fabricated, highlighting a growing issue of paper mill activity across scientific publishing.
Scientific 'spam filter' flags over 250,000 potentially fake cancer studies
A new machine learning tool has identified more than 250,000 cancer research papers that may have been produced by so-called "paper mills." Developed by QUT researcher Professor Adrian Barnett, from the School of Public Health and Social Work and Australian Center for Health Services and Innovation (AusHSI), and an international team of collaborators, the study, published in The BMJ, analyzed 2.6 million cancer studies from 1999 to 2024.
medicalxpress.com
January 30, 2026 at 1:09 AM
Wait, how many? 👀
A new paper in Nature informs us there's a new AI benchmark called Humanity’s Last Exam.
Yep, it's that same old HLE. They have submitted the paper 07 May 2025. And no, I don't know what the point of publishing it like that is either. Looks good on CVs, I guess.
Yep, it's that same old HLE. They have submitted the paper 07 May 2025. And no, I don't know what the point of publishing it like that is either. Looks good on CVs, I guess.
January 29, 2026 at 7:18 PM
A new paper in Nature informs us there's a new AI benchmark called Humanity’s Last Exam.
Yep, it's that same old HLE. They have submitted the paper 07 May 2025. And no, I don't know what the point of publishing it like that is either. Looks good on CVs, I guess.
Yep, it's that same old HLE. They have submitted the paper 07 May 2025. And no, I don't know what the point of publishing it like that is either. Looks good on CVs, I guess.
This is magic, but magic that's only available for Google AI Ultra subscribers in the U.S, so I'll just pretend it isn't interesting.
Project Genie: Experimenting with infinite, interactive worlds
Google AI Ultra subscribers in the U.S. can now try out Project Genie.
blog.google
January 29, 2026 at 5:50 PM
This is magic, but magic that's only available for Google AI Ultra subscribers in the U.S, so I'll just pretend it isn't interesting.
Now that LLMs are already solving e.g. Erdos Problems, this is very logical and interesting next step for benchmarking.
All progress is significant, as humanity's baseline is also at zero. The very best humans are estimated to have 50% chance of solving these with weeks or years of full-time work.
All progress is significant, as humanity's baseline is also at zero. The very best humans are estimated to have 50% chance of solving these with weeks or years of full-time work.
Can AI solve math research problems that have eluded human mathematicians? Our new benchmark, FrontierMath: Open Problems, is designed to help find out.
AI hasn’t solved any of these yet, but the game is young!
AI hasn’t solved any of these yet, but the game is young!
January 28, 2026 at 4:53 PM
Now that LLMs are already solving e.g. Erdos Problems, this is very logical and interesting next step for benchmarking.
All progress is significant, as humanity's baseline is also at zero. The very best humans are estimated to have 50% chance of solving these with weeks or years of full-time work.
All progress is significant, as humanity's baseline is also at zero. The very best humans are estimated to have 50% chance of solving these with weeks or years of full-time work.
"LLMs don't really understand."
Said a human who doesn't know and can't explain what that actually means.
Said a human who doesn't know and can't explain what that actually means.
"I think LLMs are just parroting their training data."
Said a human who just learned that statement from the Internet.
Said a human who just learned that statement from the Internet.
January 28, 2026 at 11:53 AM
"LLMs don't really understand."
Said a human who doesn't know and can't explain what that actually means.
Said a human who doesn't know and can't explain what that actually means.
Seems bad, but I have a solution to such problems with totalitarian government control.
Why not force the sale of TikTok USDS to some private Chinese company, like ByteDance, which has expertise running that sort of thing.
Why not force the sale of TikTok USDS to some private Chinese company, like ByteDance, which has expertise running that sort of thing.
January 27, 2026 at 11:51 PM
Seems bad, but I have a solution to such problems with totalitarian government control.
Why not force the sale of TikTok USDS to some private Chinese company, like ByteDance, which has expertise running that sort of thing.
Why not force the sale of TikTok USDS to some private Chinese company, like ByteDance, which has expertise running that sort of thing.
Gemini 3 Flash got Agentic Vision, which is cool.
But their demo app indicates it's not perfect yet.
blog.google/innovation-a...
But their demo app indicates it's not perfect yet.
blog.google/innovation-a...
January 27, 2026 at 10:15 PM
Gemini 3 Flash got Agentic Vision, which is cool.
But their demo app indicates it's not perfect yet.
blog.google/innovation-a...
But their demo app indicates it's not perfect yet.
blog.google/innovation-a...
Terence Tao gets it:
"AI is teaching us...our idea of what intelligence is is not really accurate"
"we were looking for some elusive intelligent way of of thinking and we don't see it in the tools that actually solve our goals...maybe it's actually because intelligence is not what we think it is"
"AI is teaching us...our idea of what intelligence is is not really accurate"
"we were looking for some elusive intelligent way of of thinking and we don't see it in the tools that actually solve our goals...maybe it's actually because intelligence is not what we think it is"
Can AI Prove It? Terence Tao on “Big Math” and Our Theoretical Future | The Futurology Podcast
YouTube video by Berggruen Institute
youtu.be
January 26, 2026 at 11:18 PM
Terence Tao gets it:
"AI is teaching us...our idea of what intelligence is is not really accurate"
"we were looking for some elusive intelligent way of of thinking and we don't see it in the tools that actually solve our goals...maybe it's actually because intelligence is not what we think it is"
"AI is teaching us...our idea of what intelligence is is not really accurate"
"we were looking for some elusive intelligent way of of thinking and we don't see it in the tools that actually solve our goals...maybe it's actually because intelligence is not what we think it is"
"Every few months, public sentiment either becomes convinced that AI is “hitting a wall” or becomes excited about some new breakthrough...but the truth is that behind the volatility and public speculation, there has been a smooth, unyielding increase in AI’s cognitive capabilities"
Amodei: AI will challenge us as a species
Dario Amodei — The Adolescence of Technology
Confronting and Overcoming the Risks of Powerful AI
www.darioamodei.com
January 26, 2026 at 6:38 PM
"Every few months, public sentiment either becomes convinced that AI is “hitting a wall” or becomes excited about some new breakthrough...but the truth is that behind the volatility and public speculation, there has been a smooth, unyielding increase in AI’s cognitive capabilities"
Reposted by Pekka Lund
There are many serious governments in the world.
The US is not one of them.
(There are no penguins in Greenland.)
The US is not one of them.
(There are no penguins in Greenland.)
January 23, 2026 at 11:12 PM
There are many serious governments in the world.
The US is not one of them.
(There are no penguins in Greenland.)
The US is not one of them.
(There are no penguins in Greenland.)
GPT-5.2 Pro FrontierMath Tier 4 December run score has been updated from 14/48=29.2% to 15/48=31.3% after it turned out it found an error in one of the problems.
10 of those were in the held-out set OpenAI doesn't have access to.
There's only 2 problems some other model has solved but it hasn't.
10 of those were in the held-out set OpenAI doesn't have access to.
There's only 2 problems some other model has solved but it hasn't.
New record on FrontierMath Tier 4! GPT-5.2 Pro scored 31%, a substantial jump over the previous high score of 19%. Read on for details, including comments from mathematicians.
January 23, 2026 at 10:11 PM
GPT-5.2 Pro FrontierMath Tier 4 December run score has been updated from 14/48=29.2% to 15/48=31.3% after it turned out it found an error in one of the problems.
10 of those were in the held-out set OpenAI doesn't have access to.
There's only 2 problems some other model has solved but it hasn't.
10 of those were in the held-out set OpenAI doesn't have access to.
There's only 2 problems some other model has solved but it hasn't.
Interesting:
"other known psychedelic compounds also usually produce idiosyncratic trips that vary not only from person to person but also from one experience to the next within the same individual. With L. asiatica, though, "the perception of little people is very reliably and repeatedly reported"
"other known psychedelic compounds also usually produce idiosyncratic trips that vary not only from person to person but also from one experience to the next within the same individual. With L. asiatica, though, "the perception of little people is very reliably and repeatedly reported"
the idea of hallucinating dozens of elves or whatever is terrifying to me www.bbc.com/future/artic...
'They saw them on their dishes when eating': The mushroom making people hallucinate dozens of tiny humans
Only recently described by science, the mysterious mushrooms are found in different parts of the world, but they give people the same exact visions.
www.bbc.com
January 23, 2026 at 2:44 PM
Interesting:
"other known psychedelic compounds also usually produce idiosyncratic trips that vary not only from person to person but also from one experience to the next within the same individual. With L. asiatica, though, "the perception of little people is very reliably and repeatedly reported"
"other known psychedelic compounds also usually produce idiosyncratic trips that vary not only from person to person but also from one experience to the next within the same individual. With L. asiatica, though, "the perception of little people is very reliably and repeatedly reported"
"I think LLMs are just parroting their training data."
Said a human who just learned that statement from the Internet.
Said a human who just learned that statement from the Internet.
January 23, 2026 at 1:05 AM
"I think LLMs are just parroting their training data."
Said a human who just learned that statement from the Internet.
Said a human who just learned that statement from the Internet.
Former superpower now run by kids.
"The White House...posted an altered photo of an attorney arrested after a Minnesota church protest"
"Abigail Jackson, a White House spokeswoman, mocked people questioning the image with an X post that said: “uM, eXCuSe mE??? iS tHAt DiGiTAlLy AlTeReD?!?!?!?!?!”"
"The White House...posted an altered photo of an attorney arrested after a Minnesota church protest"
"Abigail Jackson, a White House spokeswoman, mocked people questioning the image with an X post that said: “uM, eXCuSe mE??? iS tHAt DiGiTAlLy AlTeReD?!?!?!?!?!”"
Image-forensic expert Hany Farid confirmed it's a fake.
@donmoyn.bsky.social told me this moves us "closer to the Stalinesque manipulation of images that we think about with authoritarian propaganda, where you really cannot trust materials the state is putting out"
Gift link: wapo.st/4sUEnmr
@donmoyn.bsky.social told me this moves us "closer to the Stalinesque manipulation of images that we think about with authoritarian propaganda, where you really cannot trust materials the state is putting out"
Gift link: wapo.st/4sUEnmr
White House shares doctored image portraying arrested church protester in tears
The photo of an attorney arrested after a Minnesota church protest was edited to make it look like she was crying. The White House’s X post had been seen roughly 2.5 million times by Thursday afternoo...
wapo.st
January 22, 2026 at 11:44 PM
Former superpower now run by kids.
"The White House...posted an altered photo of an attorney arrested after a Minnesota church protest"
"Abigail Jackson, a White House spokeswoman, mocked people questioning the image with an X post that said: “uM, eXCuSe mE??? iS tHAt DiGiTAlLy AlTeReD?!?!?!?!?!”"
"The White House...posted an altered photo of an attorney arrested after a Minnesota church protest"
"Abigail Jackson, a White House spokeswoman, mocked people questioning the image with an X post that said: “uM, eXCuSe mE??? iS tHAt DiGiTAlLy AlTeReD?!?!?!?!?!”"
Has anyone taken a closer look at that company?
I asked Gemini to do it based on the limited information on their site, and it basically ended up agreeing with what I thought.
I asked Gemini to do it based on the limited information on their site, and it basically ended up agreeing with what I thought.
January 22, 2026 at 10:26 PM
Has anyone taken a closer look at that company?
I asked Gemini to do it based on the limited information on their site, and it basically ended up agreeing with what I thought.
I asked Gemini to do it based on the limited information on their site, and it basically ended up agreeing with what I thought.
95% of supposedly scientific discussion about consciousness distilled into one image.
January 22, 2026 at 11:57 AM
95% of supposedly scientific discussion about consciousness distilled into one image.
"However, the most creative individuals still clearly outperform even the best AI systems."
"The best" of the "current LLMs" in that study was GPT-4. Results would likely be different with actual current LLMs.
Traditional publishing is just too slow for stuff like that.
"The best" of the "current LLMs" in that study was GPT-4. Results would likely be different with actual current LLMs.
Traditional publishing is just too slow for stuff like that.
Generative AI systems now match or surpass average human creativity on certain tasks, but the highest levels of creative performance remain distinctly human, highlighting the value of human–AI collaboration. doi.org/hbkvz3
Creative talent: Has AI knocked humans out?
Are generative artificial intelligence systems such as ChatGPT truly creative? A research team led by Professor Karim Jerbi from the Department of Psychology at the Université de Montréal, and including AI pioneer Yoshua Bengio, also a professor at Université de Montréal, has just published the largest comparative study ever conducted on the creativity of large language models versus humans.
techxplore.com
January 21, 2026 at 11:52 PM
"However, the most creative individuals still clearly outperform even the best AI systems."
"The best" of the "current LLMs" in that study was GPT-4. Results would likely be different with actual current LLMs.
Traditional publishing is just too slow for stuff like that.
"The best" of the "current LLMs" in that study was GPT-4. Results would likely be different with actual current LLMs.
Traditional publishing is just too slow for stuff like that.
Reposted by Pekka Lund
🚨HOLY CRAP. An ICE whistleblower just revealed a secret memo authorizing ICE officers to break into homes without a judicial warrant, which DHS's own legal training materials say is unconstitutional!
ICE then hid the memo from the public, passing it along by word of mouth and private conversation.
ICE then hid the memo from the public, passing it along by word of mouth and private conversation.
January 21, 2026 at 10:00 PM
🚨HOLY CRAP. An ICE whistleblower just revealed a secret memo authorizing ICE officers to break into homes without a judicial warrant, which DHS's own legal training materials say is unconstitutional!
ICE then hid the memo from the public, passing it along by word of mouth and private conversation.
ICE then hid the memo from the public, passing it along by word of mouth and private conversation.
Reposted by Pekka Lund
The choice is impeachment and removal or calamity for the United States. I don't see how anybody watching Trump's speech in Davos can draw any other conclusion. He's a senile madman.
January 21, 2026 at 2:26 PM
The choice is impeachment and removal or calamity for the United States. I don't see how anybody watching Trump's speech in Davos can draw any other conclusion. He's a senile madman.
A pacifier could also work.
Honestly if Denmark/E.U./NATO offered to give him a symbolic gold crown and scepter as honorary King of Greenland in lieu of selling it he would accept in a heartbeat and we could just end this insanity
January 21, 2026 at 2:58 PM
A pacifier could also work.
Sounds like we have to wait a few more months to see the new Meta models.
"Speaking in general terms about the development cycle, Bosworth said: "There's a tremendous amount of work to do post-training" for AI, "to actually deliver the model in a way that's usable internally and by consumers.""
"Speaking in general terms about the development cycle, Bosworth said: "There's a tremendous amount of work to do post-training" for AI, "to actually deliver the model in a way that's usable internally and by consumers.""
Exclusive: Meta's new AI team delivered first key models internally this month, CTO says reut.rs/4sOaMek
Exclusive: Meta's new AI team delivered first key models internally this month, CTO says
Meta Platforms' new artificial intelligence lab has delivered its first high-profile AI models internally this month, the company's chief technology officer said on Wednesday.
reut.rs
January 21, 2026 at 2:04 PM
Sounds like we have to wait a few more months to see the new Meta models.
"Speaking in general terms about the development cycle, Bosworth said: "There's a tremendous amount of work to do post-training" for AI, "to actually deliver the model in a way that's usable internally and by consumers.""
"Speaking in general terms about the development cycle, Bosworth said: "There's a tremendous amount of work to do post-training" for AI, "to actually deliver the model in a way that's usable internally and by consumers.""
Someone had to do this.
January 20, 2026 at 1:05 AM
Someone had to do this.
Next up: A new Nobel Peace Prize competitor brought to you by the united nations of USA, Russia, Israel and North Korea.
Trump has invited Putin to join Gaza’s ‘Board of Peace’, says Kremlin – Europe live
Trump has invited Putin to join Gaza’s ‘Board of Peace’, says Kremlin – Europe live
Russia seeking to ‘clarify all the nuances’ of offer to join board meant to oversee governance and reconstruction of territory
In a nod to Trump’s efforts on Ukraine, Starmer says he recognises the US president’s role in pushing for ceasefire there – as he says “we will work closely with the United States, Ukraine and our other allies to apply pressure where it belongs: on Putin.”
In his strongest criticism of Trump yet, Starmer goes on to say:
A trade war is in no one’s interest, and my job is always to act in the UK’s national interest. Continue reading...
www.theguardian.com
January 19, 2026 at 1:40 PM
Next up: A new Nobel Peace Prize competitor brought to you by the united nations of USA, Russia, Israel and North Korea.
Reposted by Pekka Lund
The Nobel Peace Prize becoming so prestigious that wars are fought over a head of state coveting it is some real monkey paw stuff for Alfred Nobel.
January 19, 2026 at 8:47 AM
The Nobel Peace Prize becoming so prestigious that wars are fought over a head of state coveting it is some real monkey paw stuff for Alfred Nobel.
This has to be the silliest description of an attempted annexation. Even after everything that has been said about Russia and Ukraine.
"Trump, who has demanded permission from Denmark to take control of Greenland"
"Trump, who has demanded permission from Denmark to take control of Greenland"
EU readies €93bn tariffs in retaliation for Trump’s Greenland threat
Europe could also threaten to restrict US companies’ market access in run-up to crunch talks at Davos
www.ft.com
January 18, 2026 at 7:20 PM
This has to be the silliest description of an attempted annexation. Even after everything that has been said about Russia and Ukraine.
"Trump, who has demanded permission from Denmark to take control of Greenland"
"Trump, who has demanded permission from Denmark to take control of Greenland"