4 ms·
A Peter Norvig quote from yesterday's article about his transition to Stanford... "In the past, the interesting questions were around what algorithm is best fo
by pjmorris 5y ago
A Peter Norvig quote from yesterday's article about his transition to Stanford...
"In the past, the interesting questions were around what algorithm is best for doing this optimization. Now that we have a great set of algorithms and tools, the more pressing questions are human-centered: Exactly what do you want to optimize? Whose interests are you serving? Are you being fair to everyone? Is anyone being left out? Is the data you collected inclusive, or is it biased?"
[0] https://news.ycombinator.com/item?id=28833933 https://news.ycombinator.com/item?id=28833933
- jhgb 5y agoThat makes the sentence quoted by GP sound like a category error to me. Developing better algorithms and developing more useful models to run on those algorithms are not an "either/or" situation.