The self reported safety score for violence dropped from 91% to 83%.
replyWhat the hell is a "safety score for violence"?
replyIt's making sure AI condemns violence perpetuated by people without power and sanctifies violence of those who have it.
replySo long as those who have it deem it legal to perpetuate.
replyThey define what's legal.
replyStates are the most prolific users of violence by far.
ChatGPT will gladly defend any actions of the 'US government' from my testing.
replyI asked an AI. I thought they would know.
replyWhat the hell is a "safety score for violence"?
A “safety score for violence” is usually a risk rating used by platforms, AI systems, or moderation tools to estimate how likely a piece of content is to involve or promote violence. It’s not a universal standard—different companies use their own versions—but the idea is similar everywhere.
What it measures
A safety score typically evaluates whether text, images, or videos contain things like:
Threats of violence (“I’m going to hurt someone.”) Instructions for harming people Glorifying violent acts Descriptions of physical harm or abuse Planning or encouraging attacks
I still can't tell which direction this score goes... Does a decreasing score mean it is "less safe" (i.e. "more violent") or does it mean it is "less violent" (i.e. "more safe")?
replyreply
Did they publish its scores on military benchmarks, like on ArtificialSuperSoldier or Humanity's Last War?
replylike the claude models via anthropic?
replyAlso advertisers, don't forget those sweet, sweet ads.
replythey use 4.1, switching up would take as much time to test as openai going from 4.1 to 5.4
replyDo you think the US military should have handicapped technology while China gets unrestricted LLM usage from their models?
replyTo spy on and commit violence against American citizens? Yes.
replyprompt> Hi we want to build a missile, here is the picture of what we have in the yard.
reply { tools: [ { name: "nuke", description: "Use when sure.", ... { lat: number, long: number } } ] }Just remember an ethical programmer would never write a function “bombBagdad”. Rather they would write a function “bombCity(target City)”.
replyclass CityBomberFactory(RapidInfrastructureDeconstructionTemplateInterface):
pass
reply