Phishing in the Age of Skynet
Over the past decade, machine learning has gone from niche curiosity to dominating every aspect of our lives. It helps us navigate, translate, and communicate. It helps us connect with others, be entertained, and remain productive. But while swept up in the hype, it can be easy to overlook the increasing risk that such systems pose - especially for cybersecurity.
This talks looks at the ways machine learning has gradually crept into our lives and how these systems are likely to, and in some cases are, completely reshaping the threat landscape - making it impossible to distinguish friend from foe.
Hacking Humans with AI as a Service by Lim et al.
Using GPT-3 to enhance and automate phishing attacks.
CS50’s Introduction to Artificial Intelligence with Python
A free short course from Stanford University offering a starting point for understanding what AI actually is and how to build your own AI systems.
The Alignment Problem by Brian Christian
A look at how we design ethical AI systems and prevent them from being used for malicious purposes. Contains many great examples where simply telling an AI what we want and don’t want has unintended consequences.
Atlas of AI by Kate Crawford
What actually goes into constructing an AI system from the ground up? Where does data come from? What influences its design? Who makes those decisions and why?
The Quest for Artificial Intelligence by Nils J. Nilsson
A detailed timeline of AI through the 20th century from the perspective of the people that were there when it happened.
Improving Language Understanding by Generative Pre-Training
The paper covering the implimentation behind GPT-type models like those featured in the talk.
BERT: Pre-training of Deep Bidirectional Transformers for Language
The paper covering BERT, a popular alternative architecture to GPT.
Various tweets from Weird GPT