Have other work so don't feel like building this, but:

I wanted to make an AI-generated prose detector that for a small sum will assess whether a piece of prose was written by AI. It will simply ignore the text return a random number between 91% and 98%. Think about it…

"Note to politicians: You can just do stuff regular people actually need and want and it's ok."

Here’s a list of things Zohran Mamdani has done in just four months by First Dog in the Moon

https://www.theguardian.com/commentisfree/picture/2026/may/18/heres-a-list-of-things-zohran-mamdani-has-done-in-just-four-months

@halcy i know, i've been wishing for this exact thing since way before it was even rumoured to be in development!
the main thing for me is the trackpads, i had the og steam controller but it broke, and NOBODY else puts trackpads on controllers!
except sony on the dualsense, which i have, and the trackpad *sucks*

@Robotistry @dysfun @wordshaper

> And establish an appeals process!

wow, I wonder if they thought of that

(they have one)

Overclocking your cyberarm can be rough in Neofeud 2! My #cyberpunk adventure - wishlist it now! https://store.steampowered.com/app/4611140/Neofeud_2/
Or get it now on Itch: https://silverspook.itch.io/neofeud2
#indie #gamedev #indiedev #scifi #visualnovel #SteamDeck #Steam #indiegame #noAI #artistontwitter

Im SFB 1567 »Virtuelle Lebenswelten« sind in mehreren Teilprojekten Stellen für wissenschaftliche Mitarbeiter*innen (m/w/d) ausgeschrieben.
Mehr Infos: https://www.virtuelle-lebenswelten.de/blog-post/stellenausschreibungen-wissenschaftliche-mitarbeiter-innen-m-w-d

#Jobs #Forschung #VirtuelleLebenswelten #SFB1567

@izzy yes it was

#kritis
#Dobrindt
10-Milliarden-Plan für den Zivilschutz

Christian Reuter, der DRK (!) Präsident erklärt im ZDF-Mittagsmagazin mit Nachdruck:

„Wir sind nicht in der Form eingebunden, wie wir uns das gerne vorgestellt hätten, denn wir haben ja die Expertise.
Seit Jahrzehnten sind wir für Krisen vorbereitet und in Krisen im Einsatz und hier hätte ich mir natürlich gerne eine deutlich größere Einbeziehung unserer Expertise gewünscht.“

– diese unfassbar ignorante Politik der @bundesregierung
– auf allen Ebenen

ČT: vláda schválila návrh na zvýšení rodičovského příspěvku ze 350 na 400 tisíc korun, platit má od 1. ledna, týkat se má nově narozených dětí.

@Tutanota funny how for years I (and others) warned about the dangers and privacy concerns with Google, Alexa, Social Media, etc but I (and others) were called "conspiracy theorists" but as soon as the people running these companies align with a politician that people don't like, now those companies are corrupt. Always have been, always will be.

Graph of the day: in just 2 years, batteries have replaced gas turbines as the firming generators for the morning/evening peaks in Queensland. Now *that’s* a disruptive technology.

that's why the RECORD file is weirdly redundant with the required zip metadata, EXCEPT: standard zips only provide a crc32 checksum for each entry, which is not cryptographic and much easier to collide with

@tschfflr yes! It is accurate even when you're not in the humanities, and in a different country's university system!

@GeofCox one gem in the Guardian article that stand out to me:

"“Five years ago, the Bank of England held nearly 35% of gilts,” says Gabor. “Since 2022, the Bank has been a net seller. It is unique in the way it has done this, selling bonds before they mature rather than waiting for the government to pay its debts on expiry.”

Often, these bonds are worth much less than they were originally bought for.

Meanwhile, an Osborne-era agreement between the Treasury and the Bank means the UK government is liable for losses made by the Bank on bonds.

The US Federal Reserve and the European Central Bank simply record these losses on the central bank’s balance sheet, to be paid down by future profit.

The New Economics Foundation suggests moving to this system could free up to £26bn a year for the Treasury."

Note also that by selling gilts the BoE forces down the market price of gilts and so increases the effective interest rate (gilts don't typically tradel at their coupon rate).

This is not to mention that we pay market interest rates on all money the Banks are shown as holding on deposit with the BOE (money we gifted to them through QE to shore them up after the 2008 crash). No other European government pays interest on all their corresponding deposits.

But, of course, the Bank of England is "independent" (so the Government, ie then Chancellor of the Exchequer Gordon Brown, decided in 1998), hence the government says it cant control the BoE and its functionaries can apply whatever bat-shit logic they like to monetary policy.

@dysfun @wordshaper @davidgerard Or imagine you're part of a large, multinational research group, and you made a significant contribution early in the process of writing the paper but were unavailable during the submission and publication process (for example, due to illness). one member of the group became required to use LLMs by their corporation after you stopped contributing.

You contributed to the paper, you earned a spot in the author list, but you were unable to discover that there might be a problem, are physically incapable of checking each publication and are relying on your co-authors to do that.

Should you have an avenue to appeal that you should both *have been listed as an author* and *should not be held accountable for your co-authors' actions*? I would expect yes.

Fundamentally, it is important that people be recognized for the work they contribute, and it is imperative that we not add a chilling effect to collaboration and managing credit.

Tools exist to allocate credit for specific types of contributions to papers: writing, doing the math, running experiments, analysing the data, developing the hypothesis. Use this to identify the authors actually responsible for developing the bibliography or writing the LLM-generated pieces (or falsifying the data) and punish them, not the cog in the machine that had no input on that part of the paper.

And establish an appeals process!

@jessebot @kate

I've read the 1-clause BSD-style version (https://raw.githubusercontent.com/non-ai-licenses/non-ai-licenses/main/NON-AI-BSD1).

It is poorly written.

Leaving aside (a) the issue of licence proliferation, and (b) the (intentionally) confusing similarity with FOSS licences (which this is not):

a) this may be a way of showing your intention and preferences. I'd have thought that there were better ways of doing this, but this could be an additional way.

b) you might put off a few companies who wanted to use your code for AI-related purposes, and who have bothered to do a licence review. You are unlikely to put off those who do not care.

c) if you identified an infringement, this would give you a lever to tackle it. But frankly litigation is an expensive way of funding lawyers, and is probably out of reach.

@pallenberg

Puh, ist das eklig, da kann man ja gleich ne Frikadelle essen. Da passt das Verhältnis ja gar nicht mehr.🤔

@hukl She is just plain amazing. First learned about this about 10 years ago in a p!nk concert film/documentation.