It's an intellectual rug pull. Step 1: deliver information truthful as possible for an LLM, which means that hallucinations are essentially random, and therefore "harmless." Step 2: Tune the hallucinations. Goebbels would have loved this thing.
Always going to happen - I’m glad it’s so obvious for now. The future is a worry. We need a way to see inside as well as monitoring and oversight. We need something more/better than pure LLMs. Oh! I forgot to mention royalty payments too.
I'm not certain that it makes sense to reify the notion of current models as people by suggesting that they "have" personalities, beliefs, or goals. Grok may be biased to return baseless claims of a genocide against Afrikaners similar to those espoused by people who are white supremacists, but it isn't a white supremacist any more than it is a friend, a therapist, a genius, kind, intelligent, rude...because it isn't a person. Not yet, anyway.
I agree with you, Jonah, but I would think anyone subscribed to Gary Marcus would know I was just expressing something quickly and not suggesting that Grok had human consciousness.
Dear Gary, This post of yours shows striking bias. I did not see you post on OpenAI choosing to partner with far left media LeMonde as main reference. There are many similar partnerships: Grok actually systematically chooses neocons sources on international relations issues. This is a choice of their developers - here xAI. All foundational models are aligned with the biases of their developers - cultural, political correctness, etc. It did not begin with Grok, it is a feature of all of them. Finetuning at OpenAI or Anthropic is similar, with developers "telling" their AI that such or such is appropriate. To answer your question, Grok, like all AI so far, has not lost its mind, it has constructed its own behavior in a way that is only weakly controlled by the developers, but its alignment is influenced by his developers' bias. If that bias is particularly striking to you in the exemples you give it is essentially because of your preferences - in particular your dislike of Elon Musk's behavior.
I really don't care what Grok says or does not say. I don't think you do or should either.
Here are facts:
- In 1648, the Dutch East India Company had a ship shipwrecked on the Cape staffed with German, Scandinavian and Swiss refugees from the 30 years war. They thought this land would make a good colony, and the colony was established in 1652. (For comparison, the Plymouth Colony was established in 1620. Champlain founded the first permanent colony in Canada in 1608.)
- In 1648, the people of South Africa were Khoisan, related to the current bushmen of the Kalihari. These people had been there for at least tens of thousands of years.
- The Zulu of today are a Bantu people that some historians believe originated with a southern garrison of Roman legionaries due to military culture and symbols. They originated far to the north and conquered going south for thousands of years. Shaka united them in the 18th century, establishing a Zulu nation in what is now Kwazulu-Natal. The Zulu drove other Bantu tribes in front of them displacing and killing off the Khoisan.
- The Xhosa, a group closely related to the Zulu, and the Boers, encountered each other in what is now the Eastern Cape, in the late 18th century.
- The Zulus under Shaka fought the British who took over South Africa, and also fought the Boers. Boers were displace north into Natal and other areas.
- During the Second Anglo-Boer War (1899–1902), the British operated concentration camps in the South African Republic, Orange Free State, the Colony of Natal, and the Cape Colony. 26,000-47,000 Boers, mostly women and children, died of malnutrition and disease in these camps. British slaughtered livestock, burned fields, and deliberately starved the captured family members of Boers that continued to fight.
You seem to think that only when the people have actually been mostly slaughtered that genocide is real. That is no more true than in ordinary homicide. If someone is calling for your entire family's slaughter, (i.e. "Kill the Boer") and has killed some of your cousins, they can be arrested and charged with "intent to kill."
We must also consider the case of white farmers murdering black farmers. Unfortunately, many murders are racially motivated, but they do not constitute genocide.
Afrikaners really are fleeing persecution, and those denying it probably on some level think they deserve it—a little ancestral blood justice. But the farm murders are truly not a hoax, and a popular South African political leader named Julius Malema really does gather together huge crowds to chant things like: “Shoot to kill. Kill the Boer.”
Or as The New York Times has explained it: “Right-wing commenters claim that an old anti-apartheid chant is a call to anti-white violence, but historians and the left-wing politician who embraces it say it should not be taken literally.” It’s a metaphor, silly goose.
The African National Congress, South Africa’s governing party, is furious that Afrikaners are calling for help and gaining sympathy among the American right: “What the instigators of this falsehood seek is not safety, but impunity from transformation. They flee not from persecution, but from justice, equality, and accountability for historic privilege.” Impunity from transformation. Accountability for historic privilege. I can’t believe only 59 have left! My favorite part is that at the bottom of the ANC letter is a contact for media inquiries. Reach out to: Mangaliso “Stalin” Khonza, National Communications Manager. Indeed, his nom de communications guerre is Stalin. Pray tell, Stalin: What does transformation look like here?
So, run a non-deterministic program, get strange results, go to another program.
If you’re a conspiracy obsessive you will have so much text it will make Qanon look like the Declaration of Independence. Imagine! Every time you pose a “question” you get a different answer! Why! Government and industry conspiracy!
Seriously folks, figure out how these things work. It’s not hard.
Easy repair implies easily created damage... Probably just a simple context injection (systematic, random or conditional) exploiting autoregression with short messages like « white genocide in South Africa », « kill the boer »...
1) Given that Elon Musk's relatives are South African white elites, this doesn't surprise me.
2) That said, in my testing to date, most other LLMs, like ChatGPT, lean left, either in what they refuse to answer, in their preferred wording, or in the answers they give on contentious subjects. So this is nilot new, just particularly blatant. The last one this blantant was when Google's AI image generation created "diverse" images of historical scenes that were glaringly inaccurate. But they fixed that, whereas Musk's inclinations run the other way.
It's an intellectual rug pull. Step 1: deliver information truthful as possible for an LLM, which means that hallucinations are essentially random, and therefore "harmless." Step 2: Tune the hallucinations. Goebbels would have loved this thing.
Can't wait for Artificial Genocidal Intelligence!
Coming to a theatre near you, in the not too distant future.
Not to either watch or show but to shoot up the attendees like a delusional guy with a gun.
It’s a Hofstadter-Mobius loop. Known AI issue. Stay out of airlocks around Grok.
I'm sorry, Mark, I can't do that.
Great, more echo chambers and epistemic closure.
Has Grok lost its mind and mind-melded with its owner?
Wow. This AGI is really underwhelming.
“@grok is this true?” sounds harmless — but it’s a silent capitulation of independent thought.
This is one advantage of open weight models where you can at least fully control the input and output pipeline.
Always going to happen - I’m glad it’s so obvious for now. The future is a worry. We need a way to see inside as well as monitoring and oversight. We need something more/better than pure LLMs. Oh! I forgot to mention royalty payments too.
Grok is a white supremacist, and ChatGPT's recent "upgrade" made it much worse.
https://booksireadthismonth.substack.com/p/a-recent-conversation-with-chatgpt-b01
I'm not certain that it makes sense to reify the notion of current models as people by suggesting that they "have" personalities, beliefs, or goals. Grok may be biased to return baseless claims of a genocide against Afrikaners similar to those espoused by people who are white supremacists, but it isn't a white supremacist any more than it is a friend, a therapist, a genius, kind, intelligent, rude...because it isn't a person. Not yet, anyway.
I agree with you, Jonah, but I would think anyone subscribed to Gary Marcus would know I was just expressing something quickly and not suggesting that Grok had human consciousness.
Dear Gary, This post of yours shows striking bias. I did not see you post on OpenAI choosing to partner with far left media LeMonde as main reference. There are many similar partnerships: Grok actually systematically chooses neocons sources on international relations issues. This is a choice of their developers - here xAI. All foundational models are aligned with the biases of their developers - cultural, political correctness, etc. It did not begin with Grok, it is a feature of all of them. Finetuning at OpenAI or Anthropic is similar, with developers "telling" their AI that such or such is appropriate. To answer your question, Grok, like all AI so far, has not lost its mind, it has constructed its own behavior in a way that is only weakly controlled by the developers, but its alignment is influenced by his developers' bias. If that bias is particularly striking to you in the exemples you give it is essentially because of your preferences - in particular your dislike of Elon Musk's behavior.
I really don't care what Grok says or does not say. I don't think you do or should either.
Here are facts:
- In 1648, the Dutch East India Company had a ship shipwrecked on the Cape staffed with German, Scandinavian and Swiss refugees from the 30 years war. They thought this land would make a good colony, and the colony was established in 1652. (For comparison, the Plymouth Colony was established in 1620. Champlain founded the first permanent colony in Canada in 1608.)
- In 1648, the people of South Africa were Khoisan, related to the current bushmen of the Kalihari. These people had been there for at least tens of thousands of years.
- The Zulu of today are a Bantu people that some historians believe originated with a southern garrison of Roman legionaries due to military culture and symbols. They originated far to the north and conquered going south for thousands of years. Shaka united them in the 18th century, establishing a Zulu nation in what is now Kwazulu-Natal. The Zulu drove other Bantu tribes in front of them displacing and killing off the Khoisan.
- The Xhosa, a group closely related to the Zulu, and the Boers, encountered each other in what is now the Eastern Cape, in the late 18th century.
- The Zulus under Shaka fought the British who took over South Africa, and also fought the Boers. Boers were displace north into Natal and other areas.
- During the Second Anglo-Boer War (1899–1902), the British operated concentration camps in the South African Republic, Orange Free State, the Colony of Natal, and the Cape Colony. 26,000-47,000 Boers, mostly women and children, died of malnutrition and disease in these camps. British slaughtered livestock, burned fields, and deliberately starved the captured family members of Boers that continued to fight.
_________________________________________________________________________
Are current day Boers being killed by Zulus/black South Africans?
(2017) https://www.news.com.au/finance/economy/world-economy/bury-them-alive-white-south-africans-fear-for-their-future-as-horrific-farm-attacks-escalate/news-story/3a63389a1b0066b6b0b77522c06d6476
(2018) https://www.news.com.au/finance/economy/world-economy/the-time-for-reconciliation-is-over-south-africa-votes-to-confiscate-whiteowned-without-compensation/news-story/a8a81155995b1adc1c399d3576c4c0bc
(2023) https://www.news.com.au/world/africa/farmer-killings-farmers-tortured-and-killed-in-horrific-south-africa-raids/news-story/1aae3fe47328ada3b6a3d369675877df
(2023) https://www.thesouthafrican.com/news/farm-attacks-in-sa-surge-after-eff-sings-kill-the-boer-song-breaking-news-06-august-2023/
(2023) https://voz.us/en/world/231008/3986/kill-the-boer-attacks-on-white-farmers-have-increased-in-south-africa-since-julius-malemas-controversial-rally.html
Wait a minute! Farmers are killed by criminals and thugs in South Africa, but genocide is another ball game and at a different scale?
The call to genocide has been escalating since 2010. The numbers murdered has increased.
https://antiek.afriforum.co.za/wp-content/uploads/2020/07/Farm-attacks-and-farm-murders-in-South-Africa-Analysis-of-recorded-incidents-2019.pdf
You seem to think that only when the people have actually been mostly slaughtered that genocide is real. That is no more true than in ordinary homicide. If someone is calling for your entire family's slaughter, (i.e. "Kill the Boer") and has killed some of your cousins, they can be arrested and charged with "intent to kill."
Same with genocide.
We must also consider the case of white farmers murdering black farmers. Unfortunately, many murders are racially motivated, but they do not constitute genocide.
https://www.thefp.com/p/tgif-the-emir-of-america?utm_source=twitter&utm_medium=organic-social
Afrikaners really are fleeing persecution, and those denying it probably on some level think they deserve it—a little ancestral blood justice. But the farm murders are truly not a hoax, and a popular South African political leader named Julius Malema really does gather together huge crowds to chant things like: “Shoot to kill. Kill the Boer.”
Or as The New York Times has explained it: “Right-wing commenters claim that an old anti-apartheid chant is a call to anti-white violence, but historians and the left-wing politician who embraces it say it should not be taken literally.” It’s a metaphor, silly goose.
The African National Congress, South Africa’s governing party, is furious that Afrikaners are calling for help and gaining sympathy among the American right: “What the instigators of this falsehood seek is not safety, but impunity from transformation. They flee not from persecution, but from justice, equality, and accountability for historic privilege.” Impunity from transformation. Accountability for historic privilege. I can’t believe only 59 have left! My favorite part is that at the bottom of the ANC letter is a contact for media inquiries. Reach out to: Mangaliso “Stalin” Khonza, National Communications Manager. Indeed, his nom de communications guerre is Stalin. Pray tell, Stalin: What does transformation look like here?
So, run a non-deterministic program, get strange results, go to another program.
If you’re a conspiracy obsessive you will have so much text it will make Qanon look like the Declaration of Independence. Imagine! Every time you pose a “question” you get a different answer! Why! Government and industry conspiracy!
Seriously folks, figure out how these things work. It’s not hard.
If only the Gazans could have that kind of genocide, they'd trade in a heartbeat.
Was this fixed? I’m not able to make grok fall for it, its responses are decent
yes it appears to have been fixed for now. lots of bad PR
Easy repair implies easily created damage... Probably just a simple context injection (systematic, random or conditional) exploiting autoregression with short messages like « white genocide in South Africa », « kill the boer »...
It seems I was right. https://www.msn.com/en-ca/news/technology/a-rogue-employee-was-behind-grok-s-unprompted-white-genocide-mentions/ar-AA1EUIVB
Thanks for this great piece. Inspired this observation https://open.substack.com/pub/liamprompted/p/llm-news-sources-are-uniquely-susceptible?r=ckv8e&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true
1) Given that Elon Musk's relatives are South African white elites, this doesn't surprise me.
2) That said, in my testing to date, most other LLMs, like ChatGPT, lean left, either in what they refuse to answer, in their preferred wording, or in the answers they give on contentious subjects. So this is nilot new, just particularly blatant. The last one this blantant was when Google's AI image generation created "diverse" images of historical scenes that were glaringly inaccurate. But they fixed that, whereas Musk's inclinations run the other way.