Is AI actually neutral? Is that something you've even ever thought about?
Earlier this year, Paul and Mayna attended BrightonSEO and our favourite session was an eye-opening talk on the bias within AI.
So this week, we're delighted to be joined by the expert behind that session, Azahara Corrales!
Did you know that 71% of men vs. 29% of women report having AI skills (Randstad, 2024)? Did you know that Citizens Advice found customers in areas with high Black and South Asian populations were quoted at least £280 more for car insurance?
These aren't quirks of the system, they are manifestations of a deep bias that is present in the AI systems and algorithms the world (increasingly) relies upon every day.
AI consultant and adoption advisor to leaders, Azahara Corrales, has devoted time to understanding and spreading awareness of this often overlooked risk.
For anyone working with AI (that's probably all of us), and any firm rolling out AI systems, this episode should be required listening!
Chapters
00:00 Introduction to Azahara Corrales and episode overview
00:57 Azahara explains her focus on AI bias and its origins
01:54 Why this topic captured the interest of the hosts
03:10 The importance of understanding how AI works
03:54 Where AI bias actually comes from
05:58 Disparities in AI usage among genders and age groups
08:03 The bias triangle: builders, users, data
10:26 Research findings on gender bias in language models
12:02 The deep-rooted nature of AI bias and its persistence
13:28 The impact of biased AI in everyday life and decision-making
15:36 Self-fulfilling biases in AI and solutions
17:30 Practical advice for users to mitigate bias
19:10 Regulation, regulation, regulation: the need for oversight
22:59 The long-term consequences of unregulated AI
24:07 Bias beyond gender: race, religion, and other groups
27:49 AI in financial services and personal finance
40:17 What individuals and firms can do now
41:49 Jargon Busters: Flat money