this post was submitted on 03 Dec 2023
461 points (99.1% liked)
Technology
59243 readers
3375 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each another!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
Approved Bots
founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Sorry, this needs more clarification! Do you mean "intent recognition" where some AI, trained with biased data, will assume that some brown person is upto no good? Or do you mean that they will misidentify black and brown people more often due to how cameras work? Because the latter has nothing to do with biased data.
Both in fact. Training data for things like this regularly mix up minority people. If Omar is a upstanding citizen, but gets his face mixed with Haani, known terrorist, Omar gets treated unfairly, potentially to the point of lethality.
For "intent recognition", I agree. A system trained on data of mostly black committing crimes might flag more black people with ill intent.
But for the sake of identification at security checkpoints, if a man named Omar - who has an eerie resemblance to Haani the terrorist - walks through the gates, then they probably need to do a more thorough check. If they confirm with secondary data that Omar is who he says he is, then the system needs to be retrained on more images of Omar. The bias was only that they didn't have enough images of Haani and Omar for the system to make a good enough distinction. With more training, it will probably be less biased and more accurate than a human.