Hi! 👋 Welcome to The Big Y!
This week I want to touch on something that is often overlooked when discussing AI: the impact of over-optimization.
When you order multiple packages from a place like Amazon, they optimize their solution to get you all your stuff as quickly as possible, meaning that you might receive each item in a different package at a different time, delivered by a different truck. While this is optimized for quickness and customer satisfaction, it is not taking into consideration the environmental impact of the many truck deliveries.
By only optimizing for quick delivery, it inadvertently is negatively impacting the climate. One consideration here would be to add into the value function that minimizing CO2 output is also important.
Being responsible about our AI applications means taking into account more than just the profit aspect of the AI implementation. Using an accounting framework, the triple bottom line, we can think of the three elements that we should care about: profit, people, and the planet (here is an easy explanation). The question to ask is, how does your AI application impact each of the three: profit, people & the planet?
Looking at the reflections in the eyes of GAN synthesized faces (aka deepfakes) can help us detect real from fake. Deepfakes have trouble with the smaller details like small pieces of hair, and now researchers are extracting and comparing the highlights in eyes to determine whether or not a face is fake.
Thanks for reading! Share this with a friend if you think they'd like it too. Have a great week! 😁
🎙 The Big Y Podcast: Listen on Spotify, Apple Podcasts, Stitcher, Substack