Who’s behind purili? Where is it hosted? I see no information in their website about the person/community/company. That’s intriguing. Besides this purili looks neat.
Edit: It’s hosted by HETZNER-AS Hetzner Online GmbH, DE) according to https://urlscan.io/result/01a14f47-b7de-744e-b812-88ed89f058ce/
The page load statistics looks pretty clean. 1 host, 0 cookies.
Since status-quo bias can be beneficial in some contexts, the researchers do not advocate purging it from LLMs. Instead, (…) individuals who use LLMs for advice should be aware of this bias, and engineer their prompts accordingly.
It sounds like they’re saying “you’re holding it wrong”.
What about NOT asking LLMs for advices, and relying on advices from professionals or experts on the topic?
qu’est-ce qu’on va en faire ?
La conclusion du communiqué décrit assez clairement les étapes suivantes : poursuivre les essais cliniques pour démontrer l’efficacité contre l’infection du VIH, et (re)combiner le vaccin avec d’autres pour cibler plus largement les variants du virus.
La poursuite du développement devra notamment déterminer si ces réponses immunitaires peuvent se traduire par une protection contre une infection par le VIH et si cette approche peut être intégrée dans des schémas vaccinaux plus larges, capables de cibler la grande diversité génétique du virus.
En l’état, c’est prometteur car on vois que ça stimule l’immunité. Il faudrait tester sur une large population à risque pour voir si ça évite effectivement les infections, et ensuite obtenir l’autorisation des autorités de santé.
Après, en effet, viendra la question de l’accessibilité.
Terrorism scares, and is used to justify mass surveillance, exceptions in the law to give people suspected of it less rights than people suspected other serious crimes. The terrorist label is somewhat arbitrary, often deserved, sometimes abused to go after environmental activists.
In a more perfect system, the label would have a narrow and explicit definition. Or even be removed from laws, so all serious crimes suspects get the same treatment and fundamental rights, until proven guilty.
Have medical associations developed their own alternative yet?
They could save money, Doctolib subscriptions for medical profesionnals are NOT cheap. And avoid giving patient information to a large corporation with foreign investers.
How Many R’s in ‘Strawberry’?
beehaw.org defederated from lemmy.ml
Good move from the admins.
Federating seemed reasonable at first glance. I thought interacting with lemmy.ml would be worth it to see a wider range of opinions, avoid the bubble effect. It felt like interacting with yet another bubble that hates introspection and reject all critical comments outrights. Got quickly banned from a community with the reason “nazy” after I pointed out Iran’s human right violations, in a post about Iran and human rights. In retrospect those interactions where a total waste of time.
(Space) Junk Grabber
Businesses would be wise to prepare migration Away from Meta’s walled gardens/platform.
Business associations could use their weight to coordinate and promote such move, if they wanted to protect their members from being ripped off by Meta. Not holding my breath, those groups seem oblivious to big tech’s dominance.
Duct tape could be useful as an improvised protection. It requires the door to be kept closed.
A waterproof panel right in front of the door would allow to keep using the door. Obviously it requires acquiring those ahead of time, possibly work on the door opening to ensure it fits perfectly.
A strongly anti-AI policy was added to a new file: AGENTS.md in lbmk. This is the most important change in the release.
Found it’s here https://codeberg.org/libreboot/lbmk/src/branch/master/AGENTS.md
The first fixed upstream stable releases listed for the affected series are 5.10.265, 5.15.216, 6.1.183, 6.6.152, 6.12.104, 6.18.45, 7.1.9, and 7.2. Administrators should install a fixed kernel or a vendor backport and reboot into it.
The Linux 7.1.9 release was announced on Wed, 19 Aug 2026. So this was fixed about 2 months ago. This vulnerability shouldn’t be an issue for systems that are regularily updated, or disconnected from the network.
Isn’t this the plot of a Kingsman movie?
Enshittification everywhere.
I disagree. For any given attack, the attacker has most responsibilty. Destroying food is particularily vicious. If someone breaks into a house, taking davantage of an old rusted lock, and trashes the inside, the owner shouldn’t be called stupid.
It’s certainly a good idea a harden security and replace weak locks when the number of attacks increase.
Some AI companies executives are blowing smoke up their ass.
Musk said it would be full self driving several years ago.
Ça rappelle l’époque des européennes où Léa Salamé s’était mis en retrait de l’antenne, et se faisait charrier par les collègues sur France Inter. https://www.youtube.com/watch?v=NXQowAGjLC0
It’s a bad analogy. People do not conciously decide sexual orientation, and have been having sex for millenias. And we now know abstinence is a inneffective public health mesure.
Using LLMs is a concious decision, in some cases with pressure and heavy marketing from vendors. It’s a a relatively new experimental tech, we’re already aware of multiple negatives aspects (environmental, loss of skills, syncophancy, confident but wrong answer …).
The use of any experimental tech with bad side effects should be limited to small scale experiments, until we can hopefully figure out if it can be used safely-ish. What’s happening is the opposite: companies try to maximizing usage hoping we’ll eventually sort things out.