In theory, using LLMs to summarize knowledge could produce a less biased and more comprehensive output than human-written encyclopedias.
Whether Grokipedia will meet that challenge remains to be seen. But even if it doesn't, there's opportunity for other prospective encyclopedia generators to do so.
Humans looking through sources, applying knowledge of print articles and real world experiences to sift through the data, that seems far more valuable.
The perception of bias in Wikipedia remains, and if LLMs can detect and correct for bias, then Grokipedia seems at least a theoretical win.
I'm happy with at least a set of links for further research on a topic of interest.
When grok says something factual that Elon doesn't like, he puts his thumb on the scale and changes how grok responds (see the whole South African white 'genocide' business). So why should we trust that an LLM will objectively detect bias, when the people in charge of training that LLM prefer that it regurgitate their preferred story, rather than what is objectively true?
Generally, no.
With a limited domain of verifiable facts, you could perhaps measure a degree of deviation from fact across different questions, though how you get a distance measure for not just one question but that meaningfully aggregates across multiple is slippery without getting into subjective areas. Constructing a measure of directionality would be even harder to do objectively, too.
The technology behind it doesn't matter. Show me the incentives and I'll tell you the results: Wikipedia is decentralized, Grokipedia has a single owner.
I'm yet to see conservatives bring up a single subject that Wikipedia allegedly silences out of ideology, that is not an obviously false conspiracy theory. In this, Wikipedia may appear to have a left-wing bias, but only because the modern right has gotten so divorced from reality that not relaying their propaganda feels like bias against them.
Q.E.D.
If there's a perception of bias, where is it coming from? It's clearly perception born from extreme political bias of the performers. Addressing that sort of perception by changing the content means increasing bias.
Therefore the only logical route forward to hash out incidences of perceived bias and addressing them to expose them as the bias themselves.