4 ms·
So you actually believe that Twitter's AI is smart enough to detect that this statement is critical of assisted suicide, and that the data it has been trained o
by DangerousPie 6y ago
So you actually believe that Twitter's AI is smart enough to detect that this statement is critical of assisted suicide, and that the data it has been trained on was so heavily biased in favor of assisted suicide that it ends up classifying this as "undesirable content"? That sounds incredibly unlikely to me based on what I know about these kinds of algorithms.
- tgv 6y agoWell, if you take facebook's bart-large, and ask it to label the tweet on categories such as conservative, progressive, etc., you get these scores: critical 0.887 conservative 0.387 radical 0.325 progressive 0.003 If it can do that, it can generated a "biased" evaluation.
- chiefalchemist 6y agoSmart? That's the wrong methphor. That said, yes apparently so. Else it wouldn't have "canceled" this idea.