What We’ve Been Up to This Month
Congratulations to Stuart Russell for his recently announced launch of the Center for Human-Compatible AI! Read more about his new center, which is funded, primarily, by a generous grant from the Open Philanthropy Project, along with assistance from FLI and DARPA.
Why does deep learning work so well? Max Tegmark and a student just posted a paper arguing that part of the answer is to be found not in mathematics but in physics: deep learning makes approximations that are remarkably well matched to the laws of nature. We often treat deep learning like a mysterious black box much like the brain of a child: we know that if we train it according to a certain curriculum, it will learn certain skills, but we don’t fully understand why. Such improved understanding might help make AI both capable and more robust.
In addition to the EA Global conference, which four FLI members attended, Richard Mallah and Viktoriya Krakovna have also travelled far and wide to share ideas and learn more about AI and machine learning.
Mallah participated in:
Symposium on Ethics of Autonomous Systems (SEAS Europe)
Conversations on Ethical and Social Implications of Artificial Intelligence (An IEEE TechEthics™ Event)
ICYMI: This Month’s Most Popular Articles
Note From FLI:
The FLI website is including an increasing number of op-eds. Among our objectives for the website is to inspire discussion and a sharing of ideas, and as such, we’re posting opinion pieces that we believe will help spur discussion within our community. However, these op-eds do not necessarily represent FLI’s opinions or views.
We’re also looking for more writing volunteers. If you’re interested in writing either news or opinion articles for us, please let us know.