5 ms·

Yes. And because of that, the AI guesses based on the relations it sees most often. The HN title "Google translate is biased" is misleading, but it's a fascina

Yes. And because of that, the AI guesses based on the relations it sees most often. The HN title "Google translate is biased" is misleading, but it's a fascinating example of how AI can encode bias. It probably doesn't matter if I'm using Google Translate to make hotel reservations when traveling. It matters a lot if you're going to use AI to scan resumes.

mempko 6y ago
It doesn't just encode bias but perpetuates it. So yes, the AI is biased.
- fakedang 6y ago
  It's not the AI's inherent bias as such though. It's the bias of the dataset which has more cases of women cooking and men being professors in Hungarian texts.
  - Jasper_ 6y ago
    So AI is taking over the world, and we should just accept that it has severe limitations of just parroting the training data like a Markov chain? This is the algorithm we think can solve everything?
    - csmpltn 6y ago
      > So AI is taking over the world, and we should just accept that it has severe limitations of just parroting the training data like a Markov chain? This is the algorithm we think can solve everything? "No", to both questions. These systems are a work-in-progress, constantly improving. They have value, even though they are far from perfect.
  - BoiledCabbage 6y ago
    > It's not the AI's inherent bias as such though. It's the bias of the dataset... An engineer builds a chat protocol and app on top of UDP. Chat messages between users frequently never arrive and are lost. When questioned about it, the engineer says "it's not my app that causing the message loss it's UDP that's causing it." The engineer selected the communication protocol. To avoid packet loss, I build retry logic into my app, or build on top of TCP.
    - csmpltn 6y ago
      > To avoid packet loss, I build retry logic into my app, or build on top of TCP. What's the "I used TCP instead of UDP" equivalent of a dataset that can be used to build a perfectly non-biased translation service? Your comment seems to suggest that everybody seems to deliberately choose the wrong datasets which result in racist and sexist products. What should we use instead then?
      - BoiledCabbage 6y ago
        If the underlying layer isn't reliable, build a layer on top to make it reliable. If the underlying dataset has issues, build a layer on top to address those issues.

mempko 6y ago

It doesn't just encode bias but perpetuates it. So yes, the AI is biased.

fakedang 6y ago
It's not the AI's inherent bias as such though. It's the bias of the dataset which has more cases of women cooking and men being professors in Hungarian texts.
- Jasper_ 6y ago
  So AI is taking over the world, and we should just accept that it has severe limitations of just parroting the training data like a Markov chain? This is the algorithm we think can solve everything?
  - csmpltn 6y ago
    > So AI is taking over the world, and we should just accept that it has severe limitations of just parroting the training data like a Markov chain? This is the algorithm we think can solve everything? "No", to both questions. These systems are a work-in-progress, constantly improving. They have value, even though they are far from perfect.
- BoiledCabbage 6y ago
  > It's not the AI's inherent bias as such though. It's the bias of the dataset... An engineer builds a chat protocol and app on top of UDP. Chat messages between users frequently never arrive and are lost. When questioned about it, the engineer says "it's not my app that causing the message loss it's UDP that's causing it." The engineer selected the communication protocol. To avoid packet loss, I build retry logic into my app, or build on top of TCP.
  - csmpltn 6y ago
    > To avoid packet loss, I build retry logic into my app, or build on top of TCP. What's the "I used TCP instead of UDP" equivalent of a dataset that can be used to build a perfectly non-biased translation service? Your comment seems to suggest that everybody seems to deliberately choose the wrong datasets which result in racist and sexist products. What should we use instead then?
    - BoiledCabbage 6y ago
      If the underlying layer isn't reliable, build a layer on top to make it reliable. If the underlying dataset has issues, build a layer on top to address those issues.