Impossibility Theorems in AI Value Alignment (against utility functions)

Utility functions or their equivalents (value functions, objective functions,
loss functions, reward functions, preference orderings) are a central tool in
most current machine learning systems. These mechanisms for defining goals and
guiding optimization… Read more

Read full article

Similar

Detecting Photoshopped Faces Using AI

Project page: https://peterwang512.github.io/FALdetector... (more…)

Cray Advanced Supercomputer to Assist Samsung’s Research on AI and Deep Learning

SEATTLE, Nov. 15, 2017 -- Global supercomputer leader Cray Inc. (Nasdaq: CRAY) announced that Samsung Electronics Co. Ltd. has purchased a Cray CS-Storm ac... (more…)

No, Facebook’s AI wasn’t shut down because it got out of hand

Major headlines reported scary news on the power of AI: Two bots at Facebook created their own secret language and scientists had to shut… (more…)

Music Inspired by “Gödel, Escher, Bach” Created by an AI Trained on Bach

The Eternal Golden Braid by Robert M Thomas ft London Contemporary Orchestra Click above to hear something fairly amazing: Music created by a merger of Bach, artificial intelligence, and live interaction by a human audience. It's from the new album... (more…)

A leading AI ethics researcher says she’s been fired from Google

Timnit Gebru says she’s facing retaliation for conducting research that was critical of Google. (more…)

Latest News

Next job interview could be judged by AI. Here’s how to prepare

Killport – CLI command written in Rust to kill processes

Behind the Tech podcast: Bill Gates on the rapidly evolving AI landscape