Kaspar Rosager Ludvigsen

New preprint on the use of #AI in the #judiciary: can this be reconciled with the #ruleoflaw?

I identify several concerns & argue for caution to avoid that 'access to justice' turns into mere 'access to judgments', regardless of their merits.

Link: papers.ssrn.com/sol3/papers....

🙃

Before sending drones or angry comments, kindly note this is a satirical page. And satire is the essence of Democracy.

Pre-Print server SocArXiv now sets submission standards (on AI writing) than are stricter than for most high-quality peer-reviewed journals. Odd.

Så.. Blank stemme? 😌 Vi andre slipper ikke for det alle andre steder, så det skal de studerende heller ikke!

My old transport link to work and the world. We all know the reason why it gets no coverage, but it hurts to see - also considering the age and the beauty of the building.

Agreed! Genuine campus animals are so much better than mascots

You have a campus cat??? Are we the only ones who are very strict on eliminating most sources of allergies from pets (including dogs)? At any rate, very nice.

I’ve seen several potential clients using AI LLMs to evaluate legal claims and, besides often being wrong, they reliably interpret the query to predict what answer the user *wants* to hear. So, worse than wrong.

A criminal MS in the EU! Honestly feels like reading fiction, but maybe Orban needs cash for his flight to Moscow later this year.

okay, Ill bite.
What do you think the point of reading for and writing a literature review is? The process of reading and writing is crucial for THINKING. Your ideas are shaped by all of this, offloading it to gAI, no matter how "good" you think gAI is at it defeats the purpose entirely.

🙃

!! Very nice! I know mostly about auditing/verification/calibration of other types of models, but if you keep your own records and have this much of a hand in it, it should probably reveal itself eventually! The care and procedural curiosity is what got us here, it will move us past this too.

How do you verify/authenticate the translations? Does it have a cool auditing system??

It fits the original use for these language models!! I am very excited about those - just on the language. Have been trying to scramble one together in another context, but yes!! And I speak a bit of Latin anytime I interact with students, so it is good and relevant - Roman Law is still super cool.

Though, these types of SLMs do not currently exist at all.

@velocitygravity.bsky.social ! And I agree, the lack of determination in their design is a central problem. This is why I am so much more interested in SLM systems, specifically those that have used willingly given data and have not wasted tens of thousands of peoples lives and sanity to train.

It is worse than just feeling eerie in other contexts - it feels destructive and purely a void (maybe a chatbot feeling in general?). I feel nothing like this when I work with machine-learning systems, but this void eerieness is the main emotion when I use these systems!

I get this anytime I benchmark all available systems I can get my hands on with what I teach (see my papers and PhD). And similar experiences else. It is very disappointing, and eerie.

There are rumours it is considering introducing a 'commercial research exception' for AI training. This would be disastrous. It would mean handing the life's work of British creatives to AI companies for free, to train their models on.

2/4

"Women adopt AI at rates 25% lower than men.

This isn’t a guess. It’s the finding of a Harvard Business School meta-analysis examining eighteen studies, over 140,000 participants, across multiple countries."

"Every AI Company Is Building a Different Wife"

abiawomosu.substack.com/p/they-built...

Next time my colleagues from other disciplines ask me why my citation count is so low in comparison with the norm in their discipline I’m going to say our citation cartels are really small.

Flere aftner'! Mange. Mange opgaver endnu. Og 3 andre omgange endnu. Juli burde være om hjørnet om lidt (desværre ikke).

i don't know how to put this politely but stating that your work is approved/supported by companies like open ai, google and anthropic doesn't give you the credibility you think it does.

for me, this is a clear sign that i am not interested in engaging or working with you

"Hvad bruger du din aften på?" Kopiere de samme kommentarer og karakterer ind i flere systemer - at rette opgaver burde virkelig ikke gøres tre forskellige steder.

Hideo Kojima warned my generation on our PS2s that reality was subject to perspective, not the other way around. 20 years later, false images and videos can destroy reputations and shape public opinion with very few guardrails. Thankfully, Metal Gear Solid prepared me for this age of misinformation.

"Marchmænd i civil" af C. V. Jørgensen er virkelig tidsløs. Så relevant på udgivelse som nu, apropos verdens brunskjorter og udrensningspoliti. Og så er det også bare dejligt rytmisk og lækkert guitarspil. ❤️

it is day 5 of strikes at the University of Essex to save 400 jobs, our foundation year programme and our Southend campus. yesterday the VC stopped by on the picket to express disappointment that we’re on strike. i do wonder whose fault that is?