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

XAI's Grok suddenly can't stop bringing up "white genocide" in South Africa

arstechnica.com

Post reply on HN
Print view View original

31–40 of 82 posts

Re: XAI's Grok suddenly can't stop bringing up "white genocide" in South Africa

#31

Post by Lendal »

afavour wrote: post #23
sigmaisaletter wrote: post #6

Earlier quoted context omitted.

What I am wondering about is - while Musk is as unsubtle as ever, and I guess this is a system prompt instruction - is there something like that (in more subtle ways) going on in the other big models? I don't mean big agenda-pushing things like Musk, but what keeps e.g. Meta Inc. from training Llama to be ever so slightly more friendly and sympathetic to Meta Inc, or the tech industry in general? Even an open-weights…

There’s nothing stopping them at all. But in a way that’s nothing new. On one hand it feels like the height of conspiracy theory to say that Google, Meta etc would/could tweak their product to e.g. favour a particular presidential candidate. But on the other hand it’s entirely possible. Tweak what search results people see, change the weighting of what appears in their news feed… and these companies all have incentiv…

Why wouldn't they do it? If you had a backdoor into the brains of billions of people across the world (except China), and you were a billionaire with infinite ability to morally rationalize any behavior, what would stop you?

Re: XAI's Grok suddenly can't stop bringing up "white genocide" in South Africa

#32

Post by amarcheschi »

alephnan wrote: post #17
sigmaisaletter wrote: post #6

Earlier quoted context omitted.

What I am wondering about is - while Musk is as unsubtle as ever, and I guess this is a system prompt instruction - is there something like that (in more subtle ways) going on in the other big models? I don't mean big agenda-pushing things like Musk, but what keeps e.g. Meta Inc. from training Llama to be ever so slightly more friendly and sympathetic to Meta Inc, or the tech industry in general? Even an open-weights…

Wasn't the original mission of OpenAI being open and non-profit and all of that to avoid this corruption?

I don't understand why tech Ceos still have to be believed. They will say and do whatever they deem the best choice it is in their situation for profit, be it paint a thin veil of lgbt support or remove the aforementioned thin veil. The same for, well, everything that isn't lgbt/dei related such as business choices, mission, vision (...)

Re: XAI's Grok suddenly can't stop bringing up "white genocide" in South Africa

#34

Post by 1oooqooq »

sigmaisaletter wrote: post #2

> maximally truth-seeking AI, even if that truth is sometimes at odds with what is politically correct It is certainly and undoubtedly a big coincidence that his happens to the chatbot of a white South African just when the topic is in the news again due to Trump's granting refugee status to some white South African farmers.

the refugee status is a money laundering scheme. Do you think people benefiting from apartheid and now living in walled militarized praetoria (or Lesotho) need any help traveling?

banks would ask international clients the origin of the money. but not if you are opening an account under refugee status. and then they only have to pay us tax on further income, not on fortune. all that money selling black market gems to russians will be squeak clean.

it's not just something to virtue signal to their bible belt electorate. they probably sold lot of trump coins for this deal.

Re: XAI's Grok suddenly can't stop bringing up "white genocide" in South Africa

#35

Post by awongh »

What's an actually non-tinfoil hat technical explanation of the workflow that would have to happen for this to be implemented?

Elon is known for his capricious demands, (for example reports that he made the algo promote his own tweets more) how easily feasible is it to implement, "make it so Grok stops denying white genocide"? Is it just some kind of LORA you put on top of the base Grok model? It even sounds like it could be as simple as changing the text of the system prompt.

I wonder how often this happens (to any big model) without anyone noticing.

Re: XAI's Grok suddenly can't stop bringing up "white genocide" in South Africa

#36

Post by Mashimo »

MSFT_Edging wrote: post #25
philipallstar wrote: post #12

Earlier quoted context omitted.

The prospects are terrible, as being a "pale male" is the worst sort of employee to be. Businesses are given extra money if they have high BBEEE ratings; i.e. mostly-racial quotas. Loads of money siphoned off, and not just in the usual high-corruption way via the state, but e.g. if you want to procure something you go through a black-owned procurement firm that doesn't do anything except BBEEE-wash it and charge a pr…

> It's a bit like what happened in Zimbabwe that devastated that country 20 years ago when all the white farmers were kicked out Land reform to repair the inequality from the previous apartheid state isn't exactly "kicked out". Zimbabwe became Zimbabwe from Rhodesia via a civil war that wrestled control away from the minority white government. In South Africa today, 75% of the privately owned land is held by about 8%…

> Land reform to repair the inequality from the previous apartheid state isn't exactly "kicked out".

I don't know too much about Zimbabwe. From what I read about 4,000 white farmers lost house and land, and now 40 years later they are giving compensation. Why can't you say "kicked out" for that?

I do lack the knowledge to say if it was justified or not. But I consider that a kick on the way out.

Or do you mean that what happens in SA is not to be considered getting kicked out?

Re: XAI's Grok suddenly can't stop bringing up "white genocide" in South Africa

#37

Post by luma »

awongh wrote: post #35

What's an actually non-tinfoil hat technical explanation of the workflow that would have to happen for this to be implemented? Elon is known for his capricious demands, (for example reports that he made the algo promote his own tweets more) how easily feasible is it to implement, "make it so Grok stops denying white genocide"? Is it just some kind of LORA you put on top of the base Grok model? It even sounds like it…

It's clearly a ham-handed system prompt, so it's always in context which is why it keeps getting brought up. This is the level of engineering expertise that xAI brings to the table.

Re: XAI's Grok suddenly can't stop bringing up "white genocide" in South Africa

#38

Post by mcphage »

FirmwareBurner wrote: post #15
inglor_cz wrote: post #13

Earlier quoted context omitted.

[flagged]

[flagged]

> and yet nobody here made nearly as much fuss about Google forced biases and discrimination.

Maybe you weren't here for that, but... it was kinda a big deal.

Re: XAI's Grok suddenly can't stop bringing up "white genocide" in South Africa

#39

Post by Planktonne »

alephnan wrote: post #17
sigmaisaletter wrote: post #6

Earlier quoted context omitted.

What I am wondering about is - while Musk is as unsubtle as ever, and I guess this is a system prompt instruction - is there something like that (in more subtle ways) going on in the other big models? I don't mean big agenda-pushing things like Musk, but what keeps e.g. Meta Inc. from training Llama to be ever so slightly more friendly and sympathetic to Meta Inc, or the tech industry in general? Even an open-weights…

Wasn't the original mission of OpenAI being open and non-profit and all of that to avoid this corruption?

Yes, but they were lying.

Re: XAI's Grok suddenly can't stop bringing up "white genocide" in South Africa

#40

Post by pragmatic »

awongh wrote: post #35

What's an actually non-tinfoil hat technical explanation of the workflow that would have to happen for this to be implemented? Elon is known for his capricious demands, (for example reports that he made the algo promote his own tweets more) how easily feasible is it to implement, "make it so Grok stops denying white genocide"? Is it just some kind of LORA you put on top of the base Grok model? It even sounds like it…

https://www.anthropic.com/news/golden-gate-claude

Elon trying to amp white genocide.

Post reply on HN