I think this is a good thing to point out. My main reactions are: 1. I think this work was fairly low-effort and exploratory (how could we get some quick insights using a ton of AI automation), and doesn’t have the rigor I think would be needed to draw hard conclusions. For example, the Truth Social data wasn’t very high quality. 2. The absence of positive-about-AI content on Bluesky is more stark and statistically significant, and I’m much more confident that a better analysis would turn up that same result.
I think this is a good thing to point out. My main reactions are:
1. I think this work was fairly low-effort and exploratory (how could we get some quick insights using a ton of AI automation), and doesn’t have the rigor I think would be needed to draw hard conclusions. For example, the Truth Social data wasn’t very high quality.
2. The absence of positive-about-AI content on Bluesky is more stark and statistically significant, and I’m much more confident that a better analysis would turn up that same result.