Starting this week I’ve decided to start a weekly ICYMI repository showcase blog starring a few of the repositories I’ve found this week and considered interesting.
This week it’s mainly about becoming a better me and nailing that interview with an interesting entry on how to build a safe A.I.
I like this one because it helps you look at what paths you might want to take and paints a clearer picture of the available options. Features both frontend and backend, soon to have DevOps as well. Watch this repository for updates!
Below you find a set of charts demonstrating the paths that you can take and the technologies that you would want to adopt in order to become a frontend, backend or a devops. I made these charts for an old professor of mine who wanted something to share with his college students to give them a perspective.
Brush up your knowledge on different notions and principles. Data structures and algorithms explained, handful of useful resources.
TLDR: In this blogpost, we’re going to train a neural network that is fully encrypted during training (trained on unencrypted data). The result will be a neural network with two beneficial properties. First, the neural network’s intelligence is protected from those who might want to steal it, allowing valuable AIs to be trained in insecure environments without risking theft of their intelligence. Secondly, the network can only make encrypted predictions (which presumably have no impact on the outside world because the outside world cannot understand the predictions without a secret key). This creates a valuable power imbalance between a user and a superintelligence. If the AI is homomorphically encrypted, then from it’s perspective, the entire outside world is also homomorphically encrypted. A human controls the secret key and has the option to either unlock the AI itself (releasing it on the world) or just individual predictions the AI makes (seems safer).
Not exactly a repository, but a Github Page, although @iamtrask makes the code available on his GitHub.