I'm well aware of Watson's views that got him cancelled.
I know that Grok is meant to be the "uncensored, unbiased" version of LLMs. But the training data still reflects human bias, and there is definitely some irony in using an LLM for "objectivity." I do wonder what HN thinks about this though. Whether you can prompt an LLM to reflect more balanced takes that humans could do in controversial topics (assuming the LLM is "rooted" without a biased system prompt.)
Well timonoko also posts Grok generated FORTH code that doesn't work, and then retroactively claims it was just a joke when called on it, so it's safe to assume he doesn't know what he's talking about, that he and Grok and grokipedia are always joking and making shit up, and not to take any of them seriously, because he believes what he wants to believe without fact checking, and he post Grok generated AI slop regularly.
I know that Grok is meant to be the "uncensored, unbiased" version of LLMs. But the training data still reflects human bias, and there is definitely some irony in using an LLM for "objectivity." I do wonder what HN thinks about this though. Whether you can prompt an LLM to reflect more balanced takes that humans could do in controversial topics (assuming the LLM is "rooted" without a biased system prompt.)