Stanford Seminar - How can you trust machine learning? Carlos Guestrin

[@breaktherules6035]

Excellent insights! THANK YOU for sharing!

[@weisongwen3042]

very good insights, safety is definitely one of the most difficult things to be tackle…

[@johnwig285]

We are still at the infancy stage, anyways companies shouldn't focus on sprinting by adopting complicated models. Take baby steps by using simple concepts like association, clustering, predictive modelling using decision trees.

[@dertythegrower]

The data they get usually is from humans, that's the thing... most of it is not advanced at all in Ai we have today (with exceptions like IBM watson for example)

[@fgfanta]

Can we avoid injecting identity politics in machine learning too? In the bubble of the Silicon Valley, obviously not, can't avoid it.

[@cuneyt1992]

thousands of people dying from a wolf attack? and you need to train a neural network to avoid this? now that is a real proper paper motivation.

[@__________________________6910]

lol

[@444haluk]

You don't, lol