Class Central is learner-supported. When you buy through links on our site, we may earn an affiliate commission.

YouTube

Why Do AI Models Need to Be Safe? - AI Safety and Risk Management in Generative AI

IBM Research via YouTube

Overview

Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Explore the critical importance of AI safety in this 28-minute video featuring IBM Fellow Kush Varshney, who delves into how IBM Research addresses the evolving safety risks associated with generative AI. Learn about the fundamental need for AI safety measures and understand the complex process of determining what constitutes harmful AI behavior. Discover IBM's Granite Guardian system designed to detect and mitigate AI risks, while examining the crucial distinction between alignment and steerability in AI model development. Gain insights into how AI models make decisions and evaluate the effectiveness of different steering methods for controlling AI behavior. Examine intrinsic functions for generative AI and explore the emerging concept of generative computing. Understand the lessons learned from real-world AI safety implementations and IBM's approach to responsible innovation in AI development. Conclude with a forward-looking discussion on the future directions of AI safety research and the ongoing challenges in creating safe, reliable AI systems.

Syllabus

Introduction [00:00]
Why do we need AI safety? [00:09]
Determining what is harmful [1:55]
Granite Guardian to detect risks [3:00]
Alignment vs. steerability [6:17]
How to AI models make decisions? [9:35]
Are certain steering methods more effective? [12:20]
Intrinsic functions for generative AI [16:06]
What is generative computing? [17:05]
Surprises and lessons learned [19:44]
Innovating responsibly [20:58]
What's next for AI safety research [26:08]

Taught by

IBM Research

Reviews

Start your review of Why Do AI Models Need to Be Safe? - AI Safety and Risk Management in Generative AI

Never Stop Learning.

Get personalized course recommendations, track subjects and courses with reminders, and more.

Someone learning on their laptop while sitting on the floor.