Tuomas Pernu
@tuomaspernu.bsky.social
PhD and all that. Lecturer in Philosophy of Science and Research Ethics at the University of Eastern Finland (UEF). Some things I'm responsible for:
http://www.tuomaspernu.london
http://www.tuomaspernu.london
Pinned
Tuomas Pernu
@tuomaspernu.bsky.social
· Mar 7
‘It would be seen as political’: why the Royal Society is torn over Elon Musk
Many fellows feel the billionaire has breached its code of conduct, but others say scientific neutrality is at stake
www.theguardian.com
“My worry is that ejecting Musk from the Royal Society would be seen as a political move.”
My worry is that the @royalsociety.org makes the scientific community look like a bunch of pathetic pushovers!
Are you for science or against it? It's as simple as that.
www.theguardian.com/science/2025...
My worry is that the @royalsociety.org makes the scientific community look like a bunch of pathetic pushovers!
Are you for science or against it? It's as simple as that.
www.theguardian.com/science/2025...
I just came here to say.. Happy New Year!
January 1, 2026 at 1:38 AM
I just came here to say.. Happy New Year!
Reposted by Tuomas Pernu
The destruction of knowledge and history is haphazard & breathtaking.
NASA’s Largest Library Is Closing Amid Staff and Lab Cuts.
NASA’s Largest Library Is Closing Amid Staff and Lab Cuts.
NASA’s Largest Library Is Closing Amid Staff and Lab Cuts
www.nytimes.com
December 31, 2025 at 6:46 PM
The destruction of knowledge and history is haphazard & breathtaking.
NASA’s Largest Library Is Closing Amid Staff and Lab Cuts.
NASA’s Largest Library Is Closing Amid Staff and Lab Cuts.
Reposted by Tuomas Pernu
This can't be right; does not make any sense. And I'm an old-school cyclist - I'd gladly believe all sorts of abhorrent things about e-bikes! But this can't be.. *If the data is correct, I wonder if it's due to e-bikes being popular in commuting: rides happen among traffic, during rush hours.
“On a pedal bike, the chance of dying from an injury is about three-tenths of 1 percent,” Alfrey says. On an e-bike, the data indicated, it was 11 percent.”
If those data hold up nationally, that’s shocking and will require intervention.
www.nytimes.com/2025/11/30/m...
If those data hold up nationally, that’s shocking and will require intervention.
www.nytimes.com/2025/11/30/m...
The Shocking Crash That Led One County to Reckon With the Dangers of E-Bikes
www.nytimes.com
December 29, 2025 at 8:01 PM
This can't be right; does not make any sense. And I'm an old-school cyclist - I'd gladly believe all sorts of abhorrent things about e-bikes! But this can't be.. *If the data is correct, I wonder if it's due to e-bikes being popular in commuting: rides happen among traffic, during rush hours.
Reposted by Tuomas Pernu
Great thread.
"These models optimise for likely answers, not guaranteed truth. That’s a poor match for formal proofs where every step must be correct."
This is in general the case with what we call "AI" (ie machine learning): they can be of an efficient help, if you know what you are looking for.
"These models optimise for likely answers, not guaranteed truth. That’s a poor match for formal proofs where every step must be correct."
This is in general the case with what we call "AI" (ie machine learning): they can be of an efficient help, if you know what you are looking for.
AI companies have turned their attention to maths.
Between splashy headlines and sober inspection lies a more interesting story: systems that impress and develop, yet still lean on humans.
What happens when AI starts tackling maths - and what doesn’t?
1/10
Between splashy headlines and sober inspection lies a more interesting story: systems that impress and develop, yet still lean on humans.
What happens when AI starts tackling maths - and what doesn’t?
1/10
December 29, 2025 at 2:32 PM
Great thread.
"These models optimise for likely answers, not guaranteed truth. That’s a poor match for formal proofs where every step must be correct."
This is in general the case with what we call "AI" (ie machine learning): they can be of an efficient help, if you know what you are looking for.
"These models optimise for likely answers, not guaranteed truth. That’s a poor match for formal proofs where every step must be correct."
This is in general the case with what we call "AI" (ie machine learning): they can be of an efficient help, if you know what you are looking for.
Reposted by Tuomas Pernu
Currently, more Americans get their news from social and video platforms than from traditional outlets. Soon, the content on these platforms will be nothing but AI slop. What do you think will happen?
www.theatlantic.com/technology/a...
www.theatlantic.com/technology/a...
The Awkward Adolescence of a Media Revolution
Truth is still alive on social media—but it’s not easy enough to find.
www.theatlantic.com
December 27, 2025 at 6:12 PM
Currently, more Americans get their news from social and video platforms than from traditional outlets. Soon, the content on these platforms will be nothing but AI slop. What do you think will happen?
www.theatlantic.com/technology/a...
www.theatlantic.com/technology/a...
Reposted by Tuomas Pernu
Maybe we should start speaking in terms of parasites: AI parasites. That's what we are getting infested by, parasites. And at the moment, I can't see any remedy. Any other, that is, but getting off the internet - start reading books & newspapers, meeting people. Doesn't sound all bad to me!
December 27, 2025 at 7:37 PM
Maybe we should start speaking in terms of parasites: AI parasites. That's what we are getting infested by, parasites. And at the moment, I can't see any remedy. Any other, that is, but getting off the internet - start reading books & newspapers, meeting people. Doesn't sound all bad to me!
Reposted by Tuomas Pernu
Post-doc positions:
"Academic freedom is under pressure today. This requires rescue havens of free research. ... [we] invite early career researchers, whose work is restricted due to political pressure in the USA..."
uni-freiburg.de/frias/call-f...
"Academic freedom is under pressure today. This requires rescue havens of free research. ... [we] invite early career researchers, whose work is restricted due to political pressure in the USA..."
uni-freiburg.de/frias/call-f...
Call for Applications: Early Career Rescue Fellowship – Freiburg Institute for Advanced Studies
uni-freiburg.de
December 19, 2025 at 5:14 PM
Post-doc positions:
"Academic freedom is under pressure today. This requires rescue havens of free research. ... [we] invite early career researchers, whose work is restricted due to political pressure in the USA..."
uni-freiburg.de/frias/call-f...
"Academic freedom is under pressure today. This requires rescue havens of free research. ... [we] invite early career researchers, whose work is restricted due to political pressure in the USA..."
uni-freiburg.de/frias/call-f...
Reposted by Tuomas Pernu
Anyone else feel like we’re maybe two data centres away from people believing that Rodney Dangerfield was president in-between Eisenhower and Kennedy?
December 27, 2025 at 5:27 PM
Anyone else feel like we’re maybe two data centres away from people believing that Rodney Dangerfield was president in-between Eisenhower and Kennedy?
Reposted by Tuomas Pernu
another robot highlight for 2025: man wearing humanoid mocap suit kicks himself in the balls
December 27, 2025 at 5:27 PM
another robot highlight for 2025: man wearing humanoid mocap suit kicks himself in the balls
Reposted by Tuomas Pernu
I think it's clear that soon the whole of internet will be flooded by AI slop. What will happen when we can't trust anything online? That's the question we should be asking. But I don't see this question being asked, let alone answered. I think we are heading towards a collapse.
More than 20% of videos shown to new YouTube users are ‘AI slop’, study finds
More than 20% of videos shown to new YouTube users are ‘AI slop’, study finds
Low-quality AI-generated content is now saturating social media – and generating about $117m a year, data shows
More than 20% of the videos that YouTube’s algorithm shows to new users are “AI slop” – low-quality AI-generated content designed to farm views, research has found.
The video-editing company Kapwing surveyed 15,000 of the world’s most popular YouTube channels – the top 100 in every country – and found that 278 of them contain only AI slop. Continue reading...
www.theguardian.com
December 27, 2025 at 5:42 PM
I think it's clear that soon the whole of internet will be flooded by AI slop. What will happen when we can't trust anything online? That's the question we should be asking. But I don't see this question being asked, let alone answered. I think we are heading towards a collapse.
Reposted by Tuomas Pernu
South Park called it, and it’s hilarious.
One of the creators purchased the Trump Kennedy Center URL address before the name was even changed because they know Trump’s narcissism is in the driver’s seat.
One of the creators purchased the Trump Kennedy Center URL address before the name was even changed because they know Trump’s narcissism is in the driver’s seat.
December 27, 2025 at 12:38 AM
South Park called it, and it’s hilarious.
One of the creators purchased the Trump Kennedy Center URL address before the name was even changed because they know Trump’s narcissism is in the driver’s seat.
One of the creators purchased the Trump Kennedy Center URL address before the name was even changed because they know Trump’s narcissism is in the driver’s seat.
Reposted by Tuomas Pernu
Giving what we can has implemented a fun game where you spin a globe to see how your starting point in life would compare if you were reborn today, randomly somewhere on earth.
www.givingwhatwecan.org/birth-lottery
www.givingwhatwecan.org/birth-lottery
Birth Lottery
If you were reborn today, where would you land? And how would that change your life?
www.givingwhatwecan.org
December 25, 2025 at 9:31 AM
Giving what we can has implemented a fun game where you spin a globe to see how your starting point in life would compare if you were reborn today, randomly somewhere on earth.
www.givingwhatwecan.org/birth-lottery
www.givingwhatwecan.org/birth-lottery
Reposted by Tuomas Pernu
I finally got some time to flesh out my rant about these headlines...
TL;DR: age differences ≠ aging
jenndowd.substack.com/p/does-adole... #demography #episky #medsky #neurosky
TL;DR: age differences ≠ aging
jenndowd.substack.com/p/does-adole... #demography #episky #medsky #neurosky
Does adolescence really last until age 32?
Those “turning points” in brain aging aren’t quite what you think
jenndowd.substack.com
December 23, 2025 at 4:18 PM
I finally got some time to flesh out my rant about these headlines...
TL;DR: age differences ≠ aging
jenndowd.substack.com/p/does-adole... #demography #episky #medsky #neurosky
TL;DR: age differences ≠ aging
jenndowd.substack.com/p/does-adole... #demography #episky #medsky #neurosky
Reposted by Tuomas Pernu
We've got the Darwin Award, we've got Ig Nobels. We need a prize for the most idiotic innovation. It should be called, hmm.. Trump Prize?
bsky.app/profile/scie...
bsky.app/profile/scie...
Scientists have developed a robotic hand exoskeleton that can passively train expert pianists to play faster keystrokes and overcome the ceiling effect in motor skills.
Read more in #ScienceRobotics: https://scim.ag/3N3bJil
Read more in #ScienceRobotics: https://scim.ag/3N3bJil
December 24, 2025 at 12:14 AM
We've got the Darwin Award, we've got Ig Nobels. We need a prize for the most idiotic innovation. It should be called, hmm.. Trump Prize?
bsky.app/profile/scie...
bsky.app/profile/scie...
Reposted by Tuomas Pernu
Damn, can't get to this. But I find it credible that this is something that machine learning is actually good for. But let's be careful:
- People have preached the gospel of precision medicine for decades.
- If we get better at identifying risk factors, will this automatically improve well-being?
- People have preached the gospel of precision medicine for decades.
- If we get better at identifying risk factors, will this automatically improve well-being?
More accurate and individualized health predictions will allow for preventative factors to be implemented well in advance. www.wired.com/story/data-h...
Data Holds the Key in Slowing Age-Related Illnesses
More accurate and individualized health predictions will allow for preventative factors to be implemented well in advance.
www.wired.com
December 23, 2025 at 6:34 PM
Damn, can't get to this. But I find it credible that this is something that machine learning is actually good for. But let's be careful:
- People have preached the gospel of precision medicine for decades.
- If we get better at identifying risk factors, will this automatically improve well-being?
- People have preached the gospel of precision medicine for decades.
- If we get better at identifying risk factors, will this automatically improve well-being?
"Young exceptional performers and later adult world-class performers are largely two discrete populations over time."
I do think we have a serious problem in current academia in that the system incentivises - quite intentionally - quick, narrow performance. We need to change this.
I do think we have a serious problem in current academia in that the system incentivises - quite intentionally - quick, narrow performance. We need to change this.
Really interesting research showing that while ultra-specialization in a single discipline might lead to better results early in one’s career, multi-discipline training and practice pays off big time in the long run. This applies to a range of professions from scientists to athletes and more
Recent discoveries on the acquisition of the highest levels of human performance
Scientists have long debated the origins of exceptional human achievements. This literature review summarizes recent evidence from multiple domains on the acquisition of world-class performance. We re...
www.science.org
December 21, 2025 at 2:36 PM
"Young exceptional performers and later adult world-class performers are largely two discrete populations over time."
I do think we have a serious problem in current academia in that the system incentivises - quite intentionally - quick, narrow performance. We need to change this.
I do think we have a serious problem in current academia in that the system incentivises - quite intentionally - quick, narrow performance. We need to change this.
Reposted by Tuomas Pernu
When viewing the fake article in Google scholar on my university network, there is a link to access the article via my uni's library. That link sends me to a library page that makes fake article appear real... Turns out library page is made programmatically from info on Google scholar 🤦
December 21, 2025 at 1:48 AM
When viewing the fake article in Google scholar on my university network, there is a link to access the article via my uni's library. That link sends me to a library page that makes fake article appear real... Turns out library page is made programmatically from info on Google scholar 🤦
Reposted by Tuomas Pernu
Peer-reviewed publications to suffer from Gresham's Law: bad science will drive out the good. Maybe we could do better with a fully open peer review process on a website like arXiv? Authenticated users can post comments on a discussion board where everyone can read and rate the comments.
I find this very alarming. AI is being used in explicitly prohibited ways. No doubt this will soon be the norm. The consequences of this will be dramatic. Over the holidays, the publishers and the whole of academic community are doing nothing else but working on a solution to this, right? RIGHT?!
More than half of researchers now use AI for peer review — often against guidance
A survey of 1,600 academics found that more than 50% have used artificial-intelligence tools while peer reviewing manuscripts.
www.nature.com
December 19, 2025 at 11:51 AM
Peer-reviewed publications to suffer from Gresham's Law: bad science will drive out the good. Maybe we could do better with a fully open peer review process on a website like arXiv? Authenticated users can post comments on a discussion board where everyone can read and rate the comments.
Reposted by Tuomas Pernu
i’m going to start telling students to avoid using articles from 2022–onward
I find this very alarming. AI is being used in explicitly prohibited ways. No doubt this will soon be the norm. The consequences of this will be dramatic. Over the holidays, the publishers and the whole of academic community are doing nothing else but working on a solution to this, right? RIGHT?!
More than half of researchers now use AI for peer review — often against guidance
A survey of 1,600 academics found that more than 50% have used artificial-intelligence tools while peer reviewing manuscripts.
www.nature.com
December 19, 2025 at 11:25 AM
i’m going to start telling students to avoid using articles from 2022–onward
Reposted by Tuomas Pernu
Story time. Got a paper rejected. No big deal. It happens. Except one reviewer explicitly stated they were unfamiliar with our methods and asked a GenAI to explain it. They then used this to trash the paper, oh, and the journal forbids AI use in reviews.
Livid is mild to describe my feelings.
Livid is mild to describe my feelings.
I find this very alarming. AI is being used in explicitly prohibited ways. No doubt this will soon be the norm. The consequences of this will be dramatic. Over the holidays, the publishers and the whole of academic community are doing nothing else but working on a solution to this, right? RIGHT?!
More than half of researchers now use AI for peer review — often against guidance
A survey of 1,600 academics found that more than 50% have used artificial-intelligence tools while peer reviewing manuscripts.
www.nature.com
December 19, 2025 at 1:37 AM
Story time. Got a paper rejected. No big deal. It happens. Except one reviewer explicitly stated they were unfamiliar with our methods and asked a GenAI to explain it. They then used this to trash the paper, oh, and the journal forbids AI use in reviews.
Livid is mild to describe my feelings.
Livid is mild to describe my feelings.
Reposted by Tuomas Pernu
I find this very alarming. AI is being used in explicitly prohibited ways. No doubt this will soon be the norm. The consequences of this will be dramatic. Over the holidays, the publishers and the whole of academic community are doing nothing else but working on a solution to this, right? RIGHT?!
More than half of researchers now use AI for peer review — often against guidance
A survey of 1,600 academics found that more than 50% have used artificial-intelligence tools while peer reviewing manuscripts.
www.nature.com
December 18, 2025 at 9:40 PM
I find this very alarming. AI is being used in explicitly prohibited ways. No doubt this will soon be the norm. The consequences of this will be dramatic. Over the holidays, the publishers and the whole of academic community are doing nothing else but working on a solution to this, right? RIGHT?!
Reposted by Tuomas Pernu
Fellow academic writers: to make the AI summary redundant, state your conclusion rather than your topic or question in the title and write an abstract that summarises the content.
The field biologists understand this, but it hasn't reached the rest of us. And yes, I'm guilty of not doing this too.
The field biologists understand this, but it hasn't reached the rest of us. And yes, I'm guilty of not doing this too.
December 17, 2025 at 9:48 AM
Fellow academic writers: to make the AI summary redundant, state your conclusion rather than your topic or question in the title and write an abstract that summarises the content.
The field biologists understand this, but it hasn't reached the rest of us. And yes, I'm guilty of not doing this too.
The field biologists understand this, but it hasn't reached the rest of us. And yes, I'm guilty of not doing this too.
Reposted by Tuomas Pernu
Minusta olisi hallitukselta suoraselkäistä, että lähtisivät ajamaan maamme nimen muuttamista Fingoliaksi.
December 17, 2025 at 9:36 AM
Minusta olisi hallitukselta suoraselkäistä, että lähtisivät ajamaan maamme nimen muuttamista Fingoliaksi.
I've got an idea: why don't we accompany research papers with concise pieces of text summarising the structure and the main ideas of the paper? I think this could be a thing! We could call it, I don't know, maybe "abstract" or something.
"Ask AI to summarize long sections, clarify jargon, or outline the structure. Always verify summaries against the original."
Frontiers guidance on AI in peer review for ECRs optimistically expects them to "verify summaries" of things they couldn't be bothered to read it in the first place.
Frontiers guidance on AI in peer review for ECRs optimistically expects them to "verify summaries" of things they couldn't be bothered to read it in the first place.
AI in peer review: what early-career researchers should know (and how to use it wisely)
For early-career researchers, AI has become part of everyday work, whether in drafting manuscripts, organizing ideas, or exploring literature. A survey of 1,645
www.frontiersin.org
December 17, 2025 at 8:33 AM
I've got an idea: why don't we accompany research papers with concise pieces of text summarising the structure and the main ideas of the paper? I think this could be a thing! We could call it, I don't know, maybe "abstract" or something.
Reposted by Tuomas Pernu
Just decline the peer review invitation.
What are you people even doing?
What are you people even doing?
More than half of researchers now use AI for peer review — often against guidance
A survey of 1,600 academics found that more than 50% have used artificial-intelligence tools while peer reviewing manuscripts.
www.nature.com
December 16, 2025 at 10:39 PM
Just decline the peer review invitation.
What are you people even doing?
What are you people even doing?