Google Introduces AI Red Team

The icon they designed for the 'Adversarial examples' category gave me a chuckle

https://storage.googleapis.com/gweb-uniblog-publish-prod/ima...

The challenge with these high production value teams is that where they start in the fact that they even exist is long after the question of whether developing AI can even be done ethically or securely in the first place.

Obviously there are cat out of the bag arguments that hold water, but the inequality, end of civilization, potentially catastrophic weaponization, and centralization of control issues that the development of AI raises are hardly settled debates.

Short, concise and informative. Good read!

Seems like the original name of the blog post would have been fine to keep

Ah, finally a neoliberal version of AI ethics.

Pretty standard practice at this point right?

Won’t they just fire them? Or they have already been habituated to arrive at the correct result?