Elon Musk made a post recently saying there is “far too much garbage” in what AIs are learning from, so his plan is to use Grok to rewrite all human knowledge, edit out mistakes then teach Grok only from that.
Now reading that, my first reaction was just how confident you have to be to believe your company should decide what’s true or not for everyone else. It's not regular product update if you phrase it that way.
I kept looking at his wording, trying to see if I misunderstood something.
Image is AI generated
The idea of deleting errors and adding “missing information” sounds normal when you talk about a textbook, but this is talking about everything that exists online. I know the internet is messy for sure, but I don’t see how anyone even a billionaire could really fix that in a way everyone agrees with.
He went on further to ask users to submit “divisive facts” to help retrain the bot. My view is you open a bigger can of worms doing that, because now you have to pick which divisive facts stay and which ones are too risky. i don't think it's intelligent or even sensible to allow any single group being in charge of editing the world’s knowledge. That will definitely, and I mean most definitely bring in some subjective curation of what's true.
I could see from the comments that people were quick to support and attack anything he does, but for me it’s not about hating on Musk. Like I said, I don’t like the idea of any one person or company choosing what’s true for everyone, especially when the whole point of “knowledge” is that it’s open for debate.
Let me leave a link to the article I read and I hope you will pass by it and leave your thoughts in the comments of my blog.
Reference
Elon Musk says xAI will retrain Grok: 'Far too much garbage' | Business Insider Africa https://africa.businessinsider.com/news/elon-musk-says-xai-will-retrain-grok-far-too-much-garbage/k45v842#google_vignette