3 ms·
I'm glad that Google is part of this conversation, and they're now applying tests for bias to new models that they release. (Some of their old models are pretty
by rspeer 8y ago
I'm glad that Google is part of this conversation, and they're now applying tests for bias to new models that they release. (Some of their old models are pretty awful.)
If you want to see a further example, in the form of a Jupyter notebook demonstrating how extremely straightforward NLP leads to a racist model, here's a tutorial I wrote a while ago [1]:
[1] http://blog.conceptnet.io/posts/2017/how-to-make-a-racist-ai-without-really-trying/ http://blog.conceptnet.io/posts/2017/how-to-make-a-racist-ai...
- jfasi 8y agoI really have nothing to comment except how excellent I found your writeup. Seriously, wow.
- tomcatfish 8y agoReally good write up; it is going to help explain to people how data can lead to unhelpful biases by showing a few of your graphs!
- nl 8y agoFor those who aren't aware, @rspeer has been taking this problem seriously for years. His ConceptNet NumberBatch embeddings[1] are one of the few pre-built releases which attempt to fix this. [1] https://github.com/commonsense/conceptnet-numberbatch https://github.com/commonsense/conceptnet-numberbatch