Quick PSA for anyone who's still on LinkedIn: the site's stealing your data to train the slop machines
BlueMonday1984
joined 2 years ago
Nah, its completely on-topic - "bad tech hurting people" has become a major theme of our current times, a bad glucose monitor hurting diabetics is completely expected
The paper is 94 pages, but if you read through, they openly admit they’ve got nothing. Section 3.1, “Covert actions as a proxy for scheming”, admits directly:
Current frontier models likely lack the sophisticated awareness and goal-directedness required for competent and concerning scheming.
If those promptfondlers could read, they'd be very relieved.
I bet these guys are haunted by the malevolent artificial intelligence power of thermostats. It switched itself on!!
Considering how chatbot abuse causes psychosis and Apollo are AI doomsday cultists trying to convince themselves chatbots are malevolent, I genuinely think they might have gotten to that point.
They've already extruded a fake semblance of humanity for the government corruption machine, might as well extrude some flimsy botshit to "justify" its unjustifiable existence
A new artificially-extruded minister has given its inaugural address to Albania's parliament, defending its role as "not here to replace people, but to help them".
You exist to replace would-be resistance with a digital Quisling, shut the fuck up
"Let me remind you, the real danger to constitutions has never been the machines but the inhumane decisions of those in power," the clanker said.
Your purpose is to perpetrate inhumane decisions and protect those in power from accountability, shut the fuck up
The AI also responded to constitutional concerns, noting that the law "speaks of duties, responsibilities, transparency, without discrimination."
"I assure you, I embody these values as rigorously as any human colleague. Perhaps even more so."
You spew lies and perpetrate bigotry with every syllable you extrude from your "corpus" of stolen humanity, shut the fuck up
LLM's ability to fake solving word problems hinges on being able to crib the answer, so using aliens from cartoons (or automatically-generating random names for objects/characters) will prove highly effective until AI corps can get the answers into their training data.
As for context breaks, those will remain highly effective against LLMs pretty much forever - successfully working around a context break requires reasoning, which LLMs are categorically incapable of doing.
Constantly and subtly twiddling with questions (ideally through automatic means) should prove effective as well - Apple got "reasoning" text extruders to flounder and fail at simple logic puzzles through such a method.
A damn good find - bringing particular attention to this passage, since it pinpoints one of the major causes of the rot:
The following doesn’t apply to everybody in technology, but it applies to enough of them: At some point STEM education was the only thing the Olds cared about because of something something Asia, and now we have a couple of generations that are highly educated on paper and comically unaware of the complexity of the world outside of WordPress plugins.
OT: Baldur Bjarnason's lamented how his webdev feed has turned to complete shit:
Between the direct and indirect support of fascism and the uncritical embrace of LLMs and the overwhelming majority of the dev sites in my feed reader have turned to an undifferentiated puddle of nonsense…
…Two years ago these feeds (I never subscribed to any of the React grifters) were all largely posts on concrete problem-solving and, y’know, useful stuff. Useful dev discourse has collapsed into a tiny handful of blogs.
Also accidentally posted in an old thread:
Hot take: If a text extruder’s winning gold medals at your contest, that’s not a sign the text extruder’s good at something, that’s a sign your contest is worthless for determining skill.
Hot take: If a text extruder's winning gold medals at your contest, that's not a sign the text extruder's good at something, that's a sign your contest is worthless for determining skill.
LLVM is having a discussion on how to handle vibe coders attacking the project, and its causing Discourse^tm^ on the red site.
A second trashfire has hit the Ruby language - RubyGems was hit with a hostile takeover from Ruby Central, seemingly to squash an attempt at putting together an official governance policy.
Between that and DHH's fascist screed, its not been a good week for Ruby.