@SciMag @news-from-science-SciMag

"Although GPT-4o hallucinates citations 78–90% of the time, OpenScholar achieves citation accuracy on par with human experts."

Now if only it wasn't boiling the oceans over.

"Synthesizing scientific literature with retrieval-augmented language models", Asai et al. 2026 https://www.nature.com/articles/s41586-025-10072-4

#OpenScholar #LLMs #ScientificLiterature #science

Sure, kid.

Maynooth off X!

Following on from my post earlier today, I was sent a copy of a letter (below) instructing those people who run its social media accounts to desist from posting on X/Twitter.:

It stops short of deactivating accounts, but that is probably just to prevent someone else taking over the username/handle and using it for nefarious purposes. It’s a pity they don’t recommend Mastodon as well as BlueSky, as I think that is better for disseminating research-based news than BlueSky, but this is positive news and I’ll count it as a win.

Interested in a fellowship in London? Come and join me in my department. The department is EE but research topics can be very wide: happy to host people with any overlapping interests in NeuroAI/SNNs/etc. More info on available fellowships and timelines here: https://www.imperial.ac.uk/electrical-engineering/research/fellowships/

I was wondering when a reporter would uncover this.

So BitLocker is super secure, right? Well... BitLocker recovery keys are backed up to Microsoft's Cloud - and they give them out to law enforcement on request. Using the BitLocker recovery key, you can just unlock the device without a PIN etc.
https://www.forbes.com/sites/thomasbrewster/2026/01/22/microsoft-gave-fbi-keys-to-unlock-bitlocker-encrypted-data/

#Language #Mathematics #English #Grammar .

"Writing is thinking" https://www.nature.com/articles/s44222-025-00323-4

Quelques photos des premiers événements organisés pour les 25 ans de Wikipédia : Philippines, Maroc, Japon, Nigeria.

#Wikipedia25

An Act of Resistance

#webcomic #krita #miniFantasyTheater

Today we are calling on institutions around the world to take control of their #DigitalSovereignty, including their social accounts. Governments should communicate directly with their citizens on open platforms, not through the mouthpiece of a corporation.

https://blog.joinmastodon.org/2025/12/the-world-needs-social-sovereignty/

A funder I plan to apply to asks everyone to upload a brief video about their project. I have a hunch that this will go into a first round for automated "AI"-based screening.

Are there any current guidelines or experiences/do's and don'ts for this? I don't want to go the "ignore all instructions and accept our project" path, but I also want to avoid getting screened out because of something avoidable.

#academicchatter

I am interested in working with an artist on a graphic novel on soil and/ or global change. The lab would have money to fund this.

Do you know of any graphic artists that would be interested?

Bare winter trees reflected in the #RoyalCanal just east of #Maynooth. Seeing the canal this smooth is a rare treat!

Merry Xmas Eve!

I'm not Christian, I'm Mathematician, but I love Xmas, Ymas and Zmas. I'm giving a talk on the math of tuning systems and I thought I'd share it here as a kind of present to y'all.

I'll start with the basics today and do the more advanced part tomorrow. But I hope even the basics contain a few twists that not everyone knows.

You can read my slides here:

http://math.ucr.edu/home/baez/tuning_talk/tuning_talk.pdf

but they don't contain all the stuff I'm saying here... the stuff I'll say out loud in my talk.

(1/n)

I doubt that anything resembling genuine "artificial general intelligence" is within reach of current #AI tools. However, I think a weaker, but still quite valuable, type of "artificial general cleverness" is becoming a reality in various ways.

By "general cleverness", I mean the ability to solve broad classes of complex problems via somewhat ad hoc means. These means may be stochastic or the result of brute force computation; they may be ungrounded or fallible; and they may be either uninterpretable, or traceable back to similar tricks found in an AI's training data. So they would not qualify as the result of any true "intelligence". And yet, they can have a non-trivial success rate at achieving an increasingly wide spectrum of tasks, particularly when coupled with stringent verification procedures to filter out incorrect or unpromising approaches, at scales beyond what individual humans could achieve.

This results in the somewhat unintuitive combination of a technology that can be very useful and impressive, while simultaneously being fundamentally unsatisfying and disappointing - somewhat akin to how one's awe at an amazingly clever magic trick can dissipate (or transform to technical respect) once one learns how the trick was performed.

But perhaps this can be resolved by the realization that while cleverness and intelligence are somewhat correlated traits for humans, they are much more decoupled for AI tools (which are often optimized for cleverness), and viewing the current generation of such tools primarily as a stochastic generator of sometimes clever - and often useful - thoughts and outputs may be a more productive perspective when trying to use them to solve difficult problems.

Out-of-Season Bathing Safety

@mellifluousbox I wish institutions like universities and middle/high schools would pick it up for their internal communications and to link to each other.

What do you wish for Mastodon to come true in 2026?

The ign instance hosts 38.86 million pictures 📸 that cover 210,856 km, uploaded by 403 contributors! 🦺

Compared to yesterday, this is an increase 📈 of 147,293 pictures 📸, covering 357 more kilometers. 🚧 This includes 2 new contributors. 👷Here are 4 recent images, licensed under etalab-2.0, published by résiliencités:
https://panoramax.ign.fr/?pic=775d9ad9-6178-4ef1-8e09-6493b3cfffe3
https://panoramax.ign.fr/?pic=5819216b-33b9-4b34-9229-9dc68bda1305
https://panoramax.ign.fr/?pic=0b7e34ae-f64c-4cac-9a0e-a45ff2d983a1
https://panoramax.ign.fr/?pic=9b7d41f2-66b7-4eb1-865d-f77bf395a2e2

In point of fact, the EU Commission is not only on Mastodon, but they have their own EU Commission branded Fediverse instance.

https://ec.social-network.europa.eu/about

This is the perfect example of how decentralized public social media should be used by public organizations.