Genius Makers (Cade Metz)

Books · Technology

This summary was largely done for my own note-taking, sharing it just in case it adds more value to other people.

I have no affiliation whatsoever with anyone in this note. This is a summary largely taken for my own reference, and may contain errors :)

Source URL:

A lot of good ideas are ridiculed until they get blindingly obvious. When you know you’re on to something, stick with it even when others won’t. Yann LeCun, after his network for recognising digits was implemented in the real world, was then asked by AT&T to work on other more profitable projects. He refused. “I don’t care what you want. I want to work on computer vision”
You need to be a good marketer of your ideas to get them traction. Both in terms of removing previous stigma (titling papers deep learning instead of neural networks because neural networks had a stigma attached to them at the time). And to attract good talent
Find people who cover for your weaknesses. Geoff Hinton was more interested in ideas than in hardcore math or CS. So he found collaborators or graduate students who could take care of that
Ask people who aren’t believers, “what would convince you?”
“AI that plays games or works with video makes for a very good demo. Demos sell software, and sometimes companies”
Be mindful of the tribes you recruit from. Even as deep learning began to have inroads, a lot of Microsoft researchers didn’t believe in it because they harboured biases — pointing out issues like black box decision making despite its efficacy
Yoshua Bengio: “The robots of the future would need to sleep. They would need to sleep because they would need to dream”