@leeloo @Serenus @iagox86 @charlvdwalt "build something that works" is a wonderful specification. Have you ever considered a job as a project manager?

@leeloo @Serenus @iagox86 @charlvdwalt It's not a backdoor when the alternative is everyone's traffic being open to interception

@leeloo @Serenus @iagox86 @charlvdwalt Sometimes I think hackers take counterculture way too far. Who else would run the CA system?

@TindrasGrove @iagox86 @charlvdwalt You can also replace SSL with smallpox. Not having a smallpox vaccine is unlikely to lead to smallpox, because the smallpox vaccine largely eradicated smallpox.

@Mastokarl @pseudonym No, they're one and the same. Overreliance on AI is an extremely reliable indicator that someone is dumb as fuck.

@seyon @navi @charlvdwalt This is the cybersecurity equivalent of being an antivaxxer.

@iagox86 @charlvdwalt No, it was actually a really big issue which is why Google incentivized it. ISPs were injecting ads into random webpages, law enforcement and intelligence agencies made frequent use of it, and it was also a massive privacy risk. Almost every service provider has deep packet inspection and can/will monitor everything you do.

@charlvdwalt the content of the website doesn’t matter because the content is whatever an attacker wants it to be if you aren’t using SSL. Your brochureware is now malware.

As much as I love the job security, someone is going to have to stop these AI bros before they have us watering the crops with Brawndo.

After 7 years of external circumstances getting in the way, I finally managed to sit down with @jackrhysider and record a Darknet Diaries episode. Check it out here! :D

https://darknetdiaries.com/episode/158/

The AI Intelligence Paradox - An Argument For Why LLMs Lack Intelligence

https://marcushutchins.com/blog/tech/opinions/the-ai-intelligence-paradox.html

@bontchev They had actual economists

Soviet Union style planned economy, but make it so that no one involved has any idea what planning or an economy is.

Close enough, I guess

@davidtheeviloverlord It was cabbages don't eat goats, and it was hilarious

@PeterLG But AIs don't have logic, they're just following pre-determined rules, so how are they able to make rules for unseen problem.

Something about ChatGPT I'd really love to know the answer to: there used to be a really easy way to prove LLMs aren't intelligent and are just regurgitating training data. You take a common thought experiment like the wolf, the cabbage, and the goat, then change something, even a single word, to make the answer different.

The AI will either answer the thought experiment as if you'd change nothing, or go completely off the rails and spew some completely nonsensical output. If you try to do that now, even if you use variations of the thought experiment that shouldn't be in its the training data, it's able to identifies that your problem is a variant of the original and answer it.

The first part is easy, you could just write some code to match the original problem with the user's problem and if it doesn't sufficiently overlap, respond with "this is a variation of <x> problem". But the part I can't quite figure out is how they've achieved, is it can answer the variant of the problem.

I sincerely doubt LLMs have evolved any ability to think, so they aren't actually able to respond intelligently to the variant problem, but I have no idea how they managed to make it seem as if the AI is able to organically answer variants of the original problem is an intelligent-seeming way, despite lacking specific training data.

@jerry Yup, he's pretty republican. When you go so far right that you think lifelong republicans are "leftists" it's definitely time to check into the local mental hospital

@0x00string lmfao

@enobacon Conservative brain rot is so funny