Re:Kirin demos "electric salt spoon" at CES
by wenc on Saturday January 11, 2025 @12:36PM
(#42667429)
I watch this guy a lot. I noticed a few things.
He does two things that increase believability: he cites papers and studies, and he runs experiments on himself. This is good.
He does one thing that decreases believability: he only cites papers and studies that reinforces his contrarian points, and benignly ignores the rest. This is called being a YouTube influencer.
In ML terms, his content has high precision and low recall -- he optimizes for true positives but not false negatives. It's a form of confirmation bias. You can still use his data (which I assume to be correctly sourced), but you might want to be skeptical of his conclusions by seeking your own disconfirming evidence.
He does two things that increase believability: he cites papers and studies, and he runs experiments on himself. This is good.
He does one thing that decreases believability: he only cites papers and studies that reinforces his contrarian points, and benignly ignores the rest. This is called being a YouTube influencer.
In ML terms, his content has high precision and low recall -- he optimizes for true positives but not false negatives. It's a form of confirmation bias. You can still use his data (which I assume to be correctly sourced), but you might want to be skeptical of his conclusions by seeking your own disconfirming evidence.