Supervisors: Rheinallt Jones (Lead PI) and Michelle Bellingham from @sbohvm.gla.ac.uk, Dr W Peveler from
@uofgchem.bsky.social and @drcgl119.bsky.social from @uofgengineering.bsky.social
For information on the programme and how to apply visit
www.gla.ac.uk/colleges/mvl...
C'est de l'eau qui est utilisée pour le refroidissement des serveurs non ? Donc je suppose que c'est de l'eau du robinet et qu'une grande partie se vaporise ?
A little thread on how the crisis around GenAI in academia is also a crisis of confidence, and it makes me quite sad. [1/n]
It's kind of amazing that we showed we could cut child poverty in half for roughly 100 billion/year (absolute peanuts in a 7 trillion dollar budget), did it for a year, and then just stopped doing it.
Hello hive mind, based on my post yesterday about LLMs, what do we all know about the statement « AI consumes 1.5L of water for a simple yes or no »? Where does the water come from and where does it go?
Indeed
Ah yes the woke agenda *checks notes* inviting women to review because they're experts in their field.
write them like the guy that always get the funding!
I am looking for ethical alternatives including environmentally. And also starting research on the use of water for servers! I really believe that when it comes to grant writing for example, LLM are helping leveling the playing field (of course not writing your grant for you but helping you
Yes I am too far along and actually a decent LLM users that this is past my thinking. I also believe that the way it is enhancing my work- ie helping me to get quicker where I want to be could eventually being beneficial for the environment. I also still use my brains for sure. But it is also why
Ok specialists what is the best alternative to chatGPT-and name your criteria!
Universities must cut ties with OpenAI. And no, "cocaine is available on the streets, there's no reason we shouldn't dispense it in the dorms" is not a good argument.
www.nytimes.com/2025/10/28/o...
And this is also why (with the means and access they have) IF we needed metrics and as you say more than one would be helpful then we should be consulted.
I agree 💯 with all the things you are saying here. I have a Google scholar account although now I am revisiting (thank you for calling me out!) and yes if it is for you to use absolutely ok. I am just worried for institutions taking short cut and people bullying their way into positions.
when it comes to publications unintended consequences are always mentioned even when stuff is evidence based but we will let Google impose metrics on us?
Except as @lynnkamerlin.bsky.social pointed out this really depends on discipline culture which also then messes up with interdisciplinary work. My question is who asked for a new metric? Who was consulted? We know now that it is here it will be used. When we want to make meaningful change
I think we can safely refer to it as 💩 index.
Who was consulted for Google to do that? There is no existing universal rules across disciplines on ordered names on paper. Who asked Google to meddle with this? The violence of this is unique. What it means is that bad actors will be retroactively rewarded for bullying their place on papers
www.nature.com/articles/d41... look at the subtitle: which researchers welcome the effort?? Let’s just kill collaboration shall we? Let’s just make sure every paper is a battle. This is not going on the right direction
Soupçonnée d’avoir menti sous serment sur ses liens avec une lobbyiste des crèches privées, Aurore Bergé à été placée sous le statut de témoin assisté. C'était prévisible mais ça n'a pas empêché qu'elle soit renommée ministre, comme Dati. La macronie, ça ose tout.
www.lemonde.fr/politique/ar...
An image illustrating the European Union's Digital Services Act (DSA). The background is a dark purple with a binary code pattern of '0's and '1's. In the center, nine yellow icons representing DSA themes are arranged in a circular formation, similar to the stars on the EU flag. These icons include a bell, a locked padlock, a crossed-out 'AD' symbol indicating a restriction on advertising, a magnifying glass over a document for transparency, a shielded checkmark for safety, and a crossed-out symbol likely relating to content moderation or tracking. Below this visual representation, the text in large white letters states DIGITAL SERVICES ACT, followed by a smaller line of text that reads "TikTok and Meta preliminarily found in breach of their transparency obligations." A small European Commission logo is located in the bottom right corner.
Preliminary findings:
🚫TikTok & Meta failed to grant researchers adequate access to public data.
🚫Meta failed to provide simple ways to notify illegal content and to allow users to challenge moderation decisions.
This is a duty, not a choice.
link.europa.eu/JYQrBX
It’s #BlackHistoryMonth in the UK, so please afford me this statement:
It’s apparently unprofessional to not wear *smart* clothes to a conference…but it’s completely professional to use the word “coloured” when referring to Black and brown people…
Je regarde la carte et jaune veut dire “Au moins un polluant quantifié sans dépassement des limites de qualité” alors que vert veut dire aucun polluant quantifié. Vert ne veut pas dire qu’ils ne sont pas là mais
qu’ils ne sont soit pas là soit sous la limite de détection de la méthode de mesure(plus probable). Il ne faudrait donc pas séparer les deux. C’est exactement pour ça que la communication de tels chiffres est compliquée.
Je regarde la carte et jaune veut dire “Au moins un polluant quantifié sans dépassement des limites de qualité” alors que vert veut dire aucun polluant quantifié. Vert ne veut pas dire qu’ils ne sont pas là mais
A little busy tonight 😊. Will get back to the chat tomorrow
Which is why I am saying what we submit almost doesn’t matter (because I think most of us can tell a 2* from a 3/4*). Maybe AI could help more for submission selection in that sense-
I think panels do what they are asked- and work really hard the problem is what I have experienced during selection especially in “General Engineering” where submissions as so broad that reverting to JIFs to decide what is 3* and 4*
Very important and uplifting perspective on REF
There's been lots on this already. I think it would work in some circumstances but the submitting units that would be most disadvantaged would be the smaller, generally less well-funded. Panel members can spot hidden gems in a way AI wouldn't, I suspect. Might as well go back to JIFs!
