Pretty much like the cut of âDonât be evilâ rule of Google but more concrete where we are heading at.
I posted it in General News some time ago where the IDF uses AI to pick targets already. I guess OpenAI just wants a piece of that cake too.
By the way, the Chaos Communication Congress happened shortly before New Year (like every year) and brought high quality talks about this thread topic. It is not only all German but also includes English talks. You should be able to identify them by the English titles/summaries:
Iâve come to terms with the fact that I have absolutely zero interest in âAIâ as the term is being used these days. I should be all over it. I have a degree in the subject and am largely considered an expert on the topic of computers, programming, language models, and the general field of computer intelligence. I just find that I am totally uninterested in todayâs glorified chat-bots and the programs that generate âartâ and the like.
I look forward to the day when AI is actually real but I just donât feel like what we have today qualifies. Itâs artificial, for sure, but it isnât intelligent and itâs not creative.
I think there will be fields where current AI is proven to be the superior tool for the job. Unsure about it generating media, but more so at identifying patterns for low sensible/safety/security applications. Like opening the sliding doors of the gas station only for humans but not roaming bears. 
If I understand the brain correctly it is not wildly different in principle to artificial neural networks. The main difference is the way the nodes work compared to biological neurons and that connections between nodes are preset while in a brain they can grow over time. I can very well imagine these walls to break down some day.
But the main downside I see there will remain. The are very efficient (once trained) but blackboxes to the outside and themselves. They can identify the patterns of the universe but cannot explain them.
Pattern matching (whether itâs detecting a person instead of a bear or detecting planetary transits in front of a far away star) is an absolutely perfect job for a computer.
Where computers fail (and where I actually believe they will always fail) is in actually being creative. Creating something completely new, rather than derived from existing material, is something that requires a brain. If we ever figure out how to make an actual artificial brain, then yes, AI all the way home.
We donât need âAIâ on our computerâs desktop. We donât need it on our phones. I am very uncomfortable with how companies are forcing it into everything because itâs trendy.
On my side of things, I mostly hear of it for tools in material science, bioscience, engineering, and manufacturing. And even then itâs with the caveat of âwe need to first know the limit and actual shape of the technology to then make specialized toolsâ.
Itâs already known that it wonât be useful to directly find any new knowledge of the pattern themselves. Itâs basically going to give outputs where people are going to go âuh, thatâs more efficient. What a strange application of what we know. Good for usâ.
At best, there will be a few âhey, I went through what we made, because some were quite exotic, and realised that we can affine the formulation of some of what we know, neatâ.
I expect this to be the main consequence of machine learning, because there are a lot of money getting into it. Honestly, if you want a grant, just go and say âitâs for batteries / the electrical gridâ.
Dead Internet Theory is looking a lot like the Dead Internet RealityâŚ
Today the European Court of Human Rights officially forbid breaking end-to-end encryption because that technology is protecting everyone.
Sorry I have only this hard to understand original source, surely something else pops up soon. Or did and I missed it. 
The most imminent effect should be that the client-side scanning contained in the EU commisionâs âChat controlâ laws is clearly illegal now and has to be adjusted before continuing legislation.
Note that the European Court of Human Rights it not limited to the European Union, member states even today also include the United Kingdom. I would expect it enables to fight against the most recent Online Safety Bill there too.
Russia was a member state too until recently. The case that was ruled on here was against the Russian government originally and we can be glad them leaving did not stop the ruling.
@Urben youâll probably like this being a rust enthusiast.
Yeah I saw that as it was quickly shared on pretty much every Rust-related Discord. 
Maybe that speed up the market share a bit, I would not mind at all.
It is kinda ironic they push for that when every other leaked tool of the CIA/NSA to spy on people more or less make use of memory-unsafety.
Being the pioneers of new technology puts them in a strong position to find certain âinstabilitiesâ.
I remember sama talking about GPT-5, and the fact that they call this 4o makes it far scarier than it should.
In case you needed to know how bad the rebranding was, I had trouble finding this when I searched âtwitterâ and barely got what I wanted just using âXâ. Also X turned into hardcore porn so quickly I didnât even notice!
Huh I am quite sure such content already exists in the way described here on Twitter for a long time.
âŚor so I have been told. 
Anyway, another interesting analysis of the upcoming screenshot functionality of Windows 11:
Well of course it has. All Elon has done is just acknowledged it then codify it because he canât be arsed or is far too cheap to make any efforts to moderate it.
And, donât forget, he must make people think he did this so that he can take credit for it to satisfy his ego.
Chat-control is back on the table sadly with a ridiculous âcompromiseâ.
As a reminder, breaking end-to-end encryption was declared illegal by the ECHR.
The new idea form the Belgian government: It is fine if the users actively give consent into that. Oh and if they donât, they simply canât use the service.
Aah, that is what freedom of choice must feel like.
In the latest in a string of alarming developments, the Belgian government has proposed a new supposed âsolutionâ to the Chat Control (Child Sexual Abuse Regulation) deadlock in the Council. Providers of private communications services must ask people to consent to faulty AI-based scanning of their private chats, they suggest â or be banned from sharing images, videos and URLs!
Personally I switched from WhatsApp to Signal long ago due privacy concerns. Signal already declared to withdraw the service from Europe if this is put into law. I would wish I donât have to teach my family how to use obscure under-the-radar messengers.
This week we elect the European Parliament!
There are many important topics on the table, so if you live here and did not make up your mind, please do so and go voting.
Adobe casually requires you to agree on the new TOS, which include nothing less than non-exclusive rights on your art.
If you donât agree, you cannot start the software, contact support or even uninstall the software.
Looking at which year we have, a good guess is that they want to train AI tools with it? Who knows. They did not explain anything.
