Comment by xiphias2
9 hours ago
Marc Andereseen has talked about the downside of RLHF: it's a specific group of liberal low income people in California who did the rating, so AI has been leaning their culture.
I think OpenAI tried to diversify at least the location of the raters somewhat, but it's hard to diversify on every level.
Do you have any links to documentation of this? Andreesen has a definite bias as well, so I'm not about to just accept his say-so in a fit of Appeal to Authority.
(eg: "Cite?")
He was talking about it in the Lex Friedman interview after Trump was elected. And he was talking about a lot of things the Biden administration forced on Silicon Valley at that time (since then Google lost a case about one of these back-deals).
So no evidence then. Kind of like Lex touting his bona fides as a professor.
For anyone else unfamiliar with the term:
RLHF = Reinforcement Learning from Human Feedback
https://en.wikipedia.org/wiki/Reinforcement_learning_from_hu...
What do low income people have to do with it, when AI companies and research is borne out of Silicon Valley culture of rich, liberal Californians?
I'm still waiting for models based on the curt and abrasive stereotype of Eastern European programmers, as contrast to the sickeningly cheerful AIs we have today that couldn't sound more West Coast if they tried.
Low income and liberal is usually code for certain “undesirables” that conservatives tend to dislike. Better watch what LLM your kids use or they might end up speaking Spanish and listening to rap ;).
It's not about liking / disliking, but conservatives tend to prefer staying together even if it's a bad relatioship, and liberals prefer splitting by default if there are serious problems.
The syncopath style is clearly categorized as more liberal (do what you feel is good).
Eh, or grow up hating American and thinking they need to fly to Cuba to explain to the people are great communism is for them. Who knows.
> What do low income people have to do with it, when AI companies and research is borne out of Silicon Valley culture of rich, liberal Californians?
RLHF is "ask a human to score lots of LLM answers". So the claim is that the AI companies are hiring cheap (~poor) people from convenient locations (CA, since that's where the rest of the company is).
"Poor" in California means earning $80k/year, so they probably are not doing that. Africa / Indonesia / Philippines are better places to find English speaking RLHF workers.
Yes, this precisely it. There isn't going to be hard evidence to prove it though. Survey data that underpins some empirical studies have similar transparency issues too. This is far from a new problem.
If you adjust your mindset slightly when searching online, it's not hard to find communities of people looking for quick side work and this was huge during the covid lockdown era. There were people helping train LLMs for all kinds of purposes from education to customer service. Those startups quickly cashed out a few years ago and sold to the big players we have now.
I don't get why this is hard for people to believe (or remember)?
Poor people, to the billionaire, clearly are morally and ethically unsound.
https://pmc.ncbi.nlm.nih.gov/articles/PMC9533286/
Marc Andreesen should get HF on his own RL, because he's completely wrong.
This sounds like something Elon would say to make Grok seem "totally more amazeballs," except "anti-woke" Grok suffers from the same behavior
huh? this is completely inaccurate
You're absolutely right!
Talked about as in lied about it and you taking his words for gospel without verifying it? Looks just as bad as "Yes-Men" AI models.