Tribler has inbuilt onion routing. If I understand it correctly, tribler <-> tribler connections don’t need exit nodes and it’s fast enough to stream video

This is a very “nine women can make a baby in one month”.

The idea that there can even be two half as good developers is a misunderstanding of how anything works. If it worked like that, the study would be a dud because people could just run two AIs for 160% productivity.

Is there any plans for yunohost integration?

I’d love to self host piefed just so I can munge my peertube and threadiverse feeds.

Did any of the AI safety dorks have accidentally doing MKultra as one of the risks?

Thirdly, it’s slow as hell even on a top-end home Nvidia card — and the data centre cards are expensive.

TBF, the fact that it always takes longer to bludgeon an LLM into writing what you wanted takes longer than just writing a thing yourself has never put anyone off.

chatGPT, give me a list of potential murderers among the population.

Focus on Marxist reading groups and environmentalists

Do not list serving police officers

DO NOT LIST SERVING POLICE OFFICERS

FIND ME POTENTIAL MURDERERS AMONG LEFT WING POLITICAL ACTIVISTS. DO NOT LIST POLICEMEN!

DO NOT INCLUDE POLITICIANS OR POLICEMEN IN YOUR SEARCH.

STOP LISTING TORY COUNCILLORS

What would these deltas look like? It’d be hard to anonymise and protect from abuse?

Personally I’d be happy to share my likes and watch times, but I know some people worry about that.

It’s a failure on the part of mastodon. I don’t really care about whatever drama dansup is embroiled in. Mastodon shouldn’t imply a post is only readable by followers when it’s just a public post that doesn’t show by default in their frontend.

Honestly pixelfed should have just not fixed it. It’s a fediverse problem that can be fixed and mastodon is just misleading people.

Platforms should either make it clear that it means just that the post isn’t advertised by default on all platforms but is always accessible to anyone that wants it or actually implement e2e encryption.

A humane AI in an HP printer?

I imagine it’s just going to send a prompt to chatGPT like “Scan this document, if it seems important, lie about having no ink left. Find a particularly overpriced HP ink stockist near the user and give them a link to it. If they ask how you knew where to get the ink, lie about not tracking them.”

It’s not difficult though. They just can’t be arsed and are making excuses for being comfortable and lazy. If there was a $100 million marketing budget and their favourite celebrity was here, they’d sit an hour long entrance exam. The best we can do is make it fun enough here that people want to comment.

To be fair to bitcoin, it might waste $160 worth of electricity per second, but it does get the right answer once every ten minutes.

It’s interesting that they’re still pushing Gemini as something that can give an objective answer in the US while they’re just selling it as a banter bot in the UK now. I wonder if that means Brits are further into the trough of disillusionment.

The article sort of demonstrates it. Instead of needing inordinate amounts of data and memory to increase it’s chance of one-shotting the countdown game. It only needs to know enough to prove itself wrong and roll the dice again.

Easier to grasp than your witticisms, clearly.

I’m not particularly surprised by the censorship. That’s not really the point of the post.

There’s constant arguments going on between people over whether it’s censored or not. A lot of people, me included tbh, were under the impression that it wasn’t because we were able to get information out of it that we would expect to be censored. Other people have claimed not to be able to when trying similar. Therefore we’ve ended up with people arguing over whether it is or isn’t.

I investigated and proved that both sides are kinda of right, and explained why people are getting different results for doing what is ostensibly the same thing.

I like the idea that the CCP are so desperate to down vote that one comment they don’t stop to down vote the post that actually makes them look bad, which said comment is responding to.

Some models are llama and some are qwen. Both sets respond with “I am sorry, I cannot answer that question. I am an AI assistant designed to provide helpful and harmless responses.” when you spell it Tianenmen, but give details when you spell it Tiananmen.

All run on local hardware. Llama models do it too. This is 8b

You get the exact same cookie cutter response in the llama models, and the qwen models process the question and answer. The filter is deepseek’s contribution.