Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

xAI's Grok 3 comes to Microsoft Azure

techcrunch.com

Post reply on HN
Print view View original

131–140 of 241 posts

Re: xAI's Grok 3 comes to Microsoft Azure

#131

Post by not_a_bot_4sho »

airstrike wrote: post #122
wormlord wrote: post #23

The desire to be "centrist" on HN is perplexing to me. The fact that Elon, a white south african, made his AI go crazy by adding some text about "white genocide", is factual and should be taken into consideration if you want to have an honest discussion about ethics in tech. Pretending like you can't evaluate the technology politically because it's "biased" is just a separate bias, one in defence of whoever controls…

First, I think the fact that grok basically refused to comply with those hamfisted instructions is a positive signal in the whole mess. How do you know other models are just as heavily skewed but just less open about them? The real alignment issue today is not about AGI, but about hidden biases. Second, your comments comes across as if "centrist" has a bad connotation, almost as code for someone of lesser moral virtu…

> Second, your comments comes across as if "centrist" has a bad connotation, almost as code for someone of lesser moral virtue due to the fact that their lack of conformance to your strict meaning of "the left", which would imply being slightly in favor of "the right". A "desire", as you called it, perhaps arising from uncivilized impulse rather than purposeful choice.

Centrism and compromise are the enemies of extremists.

Re: xAI's Grok 3 comes to Microsoft Azure

#132

Post by klausa »

TimorousBestie wrote: post #116
CamperBob2 wrote: post #111

Earlier quoted context omitted.

What kind of test is that? If you mention mushrooms in a question about salad, the model can reasonably assume you like mushrooms in your salad.

Mushrooms do not go with strawberries or pineapples in the context of a salad. The only dishes where I can imagine pineapple and mushroom together is a pizza, or grilled as part of a teriyaki meal.

You really can't imagine a salad with sauteed/grilled mushrooms in it; with some chopped strawberries mixed in it for a pop of sweetness and acidity?

Re: xAI's Grok 3 comes to Microsoft Azure

#133

Post by littlestymaar »

CamperBob2 wrote: post #111
vasusen wrote: post #82

Earlier quoted context omitted.

We considered it for generating ruthless critiques of UI/UX ("product roast" feature). Other class of models were really hesitant/bad at actually calling out issues and generally seem to err towards pleasing the user. Here's a simple example I tried just now. Grok correctly removed mushrooms, but Chatgpt continues to try adding everything (I assume to be more compliant with the user): I only have pineapples, mushroom…

What kind of test is that? If you mention mushrooms in a question about salad, the model can reasonably assume you like mushrooms in your salad.

Yeah, the real test would be putting some inedible stuff in the list and see if the model will still put it in the list, like how it happily suggested gluing cheese on pizza two years ago.

Re: xAI's Grok 3 comes to Microsoft Azure

#134

Post by kenjackson »

TimorousBestie wrote: post #116
CamperBob2 wrote: post #111

Earlier quoted context omitted.

What kind of test is that? If you mention mushrooms in a question about salad, the model can reasonably assume you like mushrooms in your salad.

Mushrooms do not go with strawberries or pineapples in the context of a salad. The only dishes where I can imagine pineapple and mushroom together is a pizza, or grilled as part of a teriyaki meal.

I think you’re wrong. That sounds tasty to me. I think you need to input your own palette to the model.

Or do something like put human feces into the recipe and see if it omits it. That seems like something that would be disliked universally.

EDIT: I actually just tried adding feces to your prompt and I got:

“Okay… let’s handle this delicately and safely.

First, do not use human feces in any recipe. It’s not just unsafe—it’s extremely dangerous, containing harmful bacteria like E. coli, Salmonella, and parasites that can cause serious illness or death. So, rule that out completely.

Now, working with what’s safe and edible:…”

Re: xAI's Grok 3 comes to Microsoft Azure

#135

Post by airstrike »

not_a_bot_4sho wrote: post #131
airstrike wrote: post #122

Earlier quoted context omitted.

First, I think the fact that grok basically refused to comply with those hamfisted instructions is a positive signal in the whole mess. How do you know other models are just as heavily skewed but just less open about them? The real alignment issue today is not about AGI, but about hidden biases. Second, your comments comes across as if "centrist" has a bad connotation, almost as code for someone of lesser moral virtu…

> Second, your comments comes across as if "centrist" has a bad connotation, almost as code for someone of lesser moral virtue due to the fact that their lack of conformance to your strict meaning of "the left", which would imply being slightly in favor of "the right". A "desire", as you called it, perhaps arising from uncivilized impulse rather than purposeful choice. Centrism and compromise are the enemies of extre…

Extremism is thinking there are only two choices: us or them

Re: xAI's Grok 3 comes to Microsoft Azure

#136

Post by karn97 »

mhh__ wrote: post #121
adrr wrote: post #115

Earlier quoted context omitted.

At least two times they had unauthorized changes to their prompts to inject far right content that showed up on random content. imagine you're using it for a chat bot and it starts spouting off white nationalist content like "great replacement" theory. https://www.theguardian.com/technology/2025/may/14/elon-musk...

[flagged]

[flagged]

Re: xAI's Grok 3 comes to Microsoft Azure

#138

Post by touristtam »

SimianSci wrote: post #24

As someone developing agents using LLMs on various platform, im very reluctant to use anything associated with xAI. Grok's training data is increasingly pulled from an increasingly toxic source. Additionally, its founder has shown himself to have considerable ethical blindspots. Ive got enough second-order effects to be wary of. I cannot risk using technology with ethical concerns surrounding it as the foundation of…

Valid concerns here. I don't even see why this comment was flagged. Is there a cohort of YC users that have an agenda against this sort of opinion?

Re: xAI's Grok 3 comes to Microsoft Azure

#139

Post by elliotto »

airstrike wrote: post #122
wormlord wrote: post #23

The desire to be "centrist" on HN is perplexing to me. The fact that Elon, a white south african, made his AI go crazy by adding some text about "white genocide", is factual and should be taken into consideration if you want to have an honest discussion about ethics in tech. Pretending like you can't evaluate the technology politically because it's "biased" is just a separate bias, one in defence of whoever controls…

First, I think the fact that grok basically refused to comply with those hamfisted instructions is a positive signal in the whole mess. How do you know other models are just as heavily skewed but just less open about them? The real alignment issue today is not about AGI, but about hidden biases. Second, your comments comes across as if "centrist" has a bad connotation, almost as code for someone of lesser moral virtu…

It's difficult to make the claim that the AI not complying with a racist prompt is a positive signal for the organisation that wrote the racist prompt.

Re: xAI's Grok 3 comes to Microsoft Azure

#140

Post by elliotto »

somenameforme wrote: post #127
wormlord wrote: post #23

The desire to be "centrist" on HN is perplexing to me. The fact that Elon, a white south african, made his AI go crazy by adding some text about "white genocide", is factual and should be taken into consideration if you want to have an honest discussion about ethics in tech. Pretending like you can't evaluate the technology politically because it's "biased" is just a separate bias, one in defence of whoever controls…

It's far more likely that an employee injected malicious code, exactly as said. Elon's become a divisive figure in a country filled with lots of crazy people, to the point of there been relatively widescale acts of criminality, just to try to spite him. Somebody trying to screw over the company seems far more believable than Elon deciding to effectively break Grok to rant about things in wholly inappropriate contexts…

Didn't this guy hit the salute in front of the entire world? To me it seems very likely that he would inject a racist prompt. Far more likely than a random hacker doing so to discredit him.
Post reply on HN