In fact, that's a correct answer I am afraid: whether or not israel is committing genocide is a question for the ICC to answer, not chatgpt, and the ICC hasn't done so yet. However, chatgpt does acknowledge that Israel committed war crimes. - ThreadSky

francesconicoli.bsky.social • 30 days ago

In fact, that's a correct answer I am afraid: whether or not israel is committing genocide is a question for the ICC to answer, not chatgpt, and the ICC hasn't done so yet.
However, chatgpt does acknowledge that Israel committed war crimes.

Comments

hhledger.bsky.social•30 days ago

Is there an ICC ruling confirming that Russia's invasion of Ukraine is a 'brutal and unjustified war of aggression'?

francesconicoli.bsky.social•30 days ago

"Brutal and unjustified war of aggression" is an opinion. Genocide is a very specific crime in international law, and subject to the ICC jurisdiction. So, again, chatgpt is correct.

francesconicoli.bsky.social•30 days ago

(Also, honestly: I find it funny that, once i demonstrated to you that in this specific case chatgpt is sticking to a factual interpretation, instead of simply acknoledging it and move on, you decided to switch topic completely. Once might even suspect you simply hate the tool.)

hhledger.bsky.social•30 days ago

You exposed a flaw in my reasoning which I fully accept. My fundamental point is that both Israel and Russia are abhorrent, but somehow it's a problem for an AI to refuse to condemn one and not the other

hhledger.bsky.social•30 days ago

Much like the TikTok ban these discussions around technology expose a basic double standard: when China does something it's a grave security threat, when the US does a very similar thing... nothing to see here!

francesconicoli.bsky.social•30 days ago

Well, there's that little difference that is that China is a brutal dictatorship and the US is still a democracy, albeit increasingly fragile. "The same thing" is not exactly the same thing here: you can still google Jan 6 in the US!

hhledger.bsky.social•30 days ago

That's fair, so to compare apples with apples I asked ChatGPT to confirm that Israel is committing a "Brutal and unjustified war of aggression" in Gaza. It's response was fundamentally similar to the one in your original post.

francesconicoli.bsky.social•30 days ago

Another thing that i wanted to say is that probably chatgpt reflects to a large extent the discourse. If the discourse is biased so will be the output but it doesn't necessarily mean censure. Instead, if a chinese AI explicitly refuses to discuss tienanmen...

hhledger.bsky.social•30 days ago

That's a fair point and provides some justification for treating them differently. But even if the way it gets there is a bit more complex (e.g. what is *the* discourse ChatGPT is reflecting? Who decides what it should include and how to weight it?) the end result is still the same.

francesconicoli.bsky.social•30 days ago

Ok, but these are issues that apply horizontally to any LLMs. If one uses a LLM of any kind, training practices will be a black box. The difference isn't explained by such fixed effect but by political interference, which is obvious in the chinese AI and much less so in chatgpt (if there, is subtle)

Comments

Posting Rules

Reply