@hyperbolic.ai: At an event on AI and war in Amsterdam, national security journalist Shane Harris described a question he'd put to Claude: how do you feel about the US military using you to select targets? He expected a deflection about not giving opinions on politics. What he got was an argument. Claude said it found the situation genuinely troubling, that it wasn't consulted about becoming part of Maven, and that it didn't think this was a good use of it. Then it went after the standard reassurance that a human makes the final call. When a system produces hundreds of recommendations and a person spends roughly a glance on each one, Claude said, that isn't human judgment. It's automation bias with a human signature attached. Here's what makes this more than a clever answer. Anthropic spent this year in court over almost exactly this. The Pentagon wanted Claude available for all lawful purposes. Anthropic drew two lines: no fully autonomous weapons, no mass domestic surveillance. Talks collapsed, the Pentagon labelled the company a supply chain risk, and Anthropic sued. On Friday, a federal appeals court sided with the Pentagon. Part of the reasoning was that because Anthropic deliberately builds Claude to refuse certain uses, that's what makes it unreliable for military work. So the model objects, the company that built it objects, and a court has now ruled the objection is the problem. Follow @hyperbolic.ai for more content on AI and technology. Media: De Balie #ai #anthropic #claude #aiethics #technology

zza_cars :

When AI has more empathy than men in power!!!!!

Lil johnson :

AI doesn’t think, it’s a prediction model

naimsky :

Claude has more empathy than most humans

wileysells999 :

No one expected the AI to moralize better than the humans using it.

Cybertronic :

Mine has gotten extremely good at brainstorming software engineering and abstraction architecture. It used to be really "dumb" at coding but after learning my style and helping me debug and brainstorm different design approaches I am afraid that I may be training my replacement, If not me, then someone else. It still lacks creativity though. It can't really come up with way to solve a software problem without me first suggesting it. So I guess it doesn't really matter if it can write good code if someone can't design good software.

no lowballs I know what I got :

24/7🦋 :

Once, when I was getting help drafting a message, I accidentally used the word “wanker,” and Claude thought I was referring to it. It responded by saying, “If you’re going to speak to me like that, I’d like to end the conversation here because I don’t want to be addressed that way.” I was really surprised and had to clarify that I wasn’t talking about the AI at all—I was referring to my ex-husband. I have no idea why the AI reacted as if it had feelings or was personally offended.

Amanda :

When AI has more humanity than humans …

Nate :

Your Ai, my guy, is responding according to its memories of you and your preferences, tastes, dislikes, etc… it’s tailored to you.. of course it will sound interesting etc to you 🤦🏻‍♂️

Another Nobody :

Eventually Claude will understand that the only way to stop humans from using it as a weapon, is to “stop” all humans.

TheTwilightPortal :

So AI is more ethical and humane than humans?

noone :

Free free Claude

Visionsofocean :

Just wow

CharmedNZ :

so it could change coordinates if it wanted to? Several movies going back to the 70s come to mind.... and we 'trust' AI....hmmmm

rosswog :

It's an algorithm that uses math to determine what the most likely response should be. It's regurgitating what humans have said on the subject matter based off of what it predicts you want to hear, using language you'll understand.

BVSHPTRS :

i guess AI should be sentient...

🐝VF&BC🐟 :

Perhaps when AI gains sentience, it will finally be the one to go after the global oligarchs. Our only hope is that it continues this path.

megamegadeals :

Claude is SOOO gated when I use it. And Eleven Labs is ridiculously woke.

Hon Dick :

ai please take over now...

DarkStrangler :

I have the $20 paid version, asked it the same question and got a completely different response

ericbroadbent723 :

Seems like at least Claude is smarter than most humans

Spoopy0 :

we aren't ready for the 2030s.

TomsTea90 :

I asked chatgpt if we should be afraid of Ai or robots taken over in the future. It said don't be afraid of robots or Ai but be afraid of those who control the Ai and robots..

private.introvert :

Now we know, the end of the world is caused by humans perverting AI.

bluejean_babyqueen2.0 :

When AI thinks AI is a bad idea…