• 0 Posts
  • 13 Comments
Joined 1 year ago
cake
Cake day: June 30th, 2023

help-circle





  • it was never trained to do insider trading, or any kind of trading. it was trained to predict the most likely next word given a bunch of previous words as input, then trained to not do that when the next word would be racist/destructive/etc. it turns out that’s super versatile, and can be used to approximate a lot of other functions, like trading on the stock market.

    as for sources of information, kinda the big problem with it is how unselective openai were when picking training data. they just loaded all of reddit and wikipedia into it, then dumped a ton of other random shit in there as well.

    what i’m getting at is chatgpt is really powerful (duh) but it wasn’t created with nearly the intentionality most people think it was, and it doesn’t have a lot of the power that people think it does.





  • this is misleading. if you read the study, you’ll see the aggregate miss rate is much worse for dark skinned people and children, but that’s weighted by the models that haven’t been trained specifically for use with pedestrian detection. the pedestrian detection models, i.e. the ones people actually are going to use, were within 1% as accurate for skin tone, and were about 10% worse at detecting kids. since kids are already harder to see for human drivers, that seems much more similar to human performance.

    the other thing it doesn’t mention is how bad they are at detecting people in general. the best one misses people roughly 5% of the time, which seems pretty high for something that’s supposed to be driving a car