Here is the full transcript and summary of Steven Zapata’s talk titled “The Problem With AI-Generated Art” at TEDxBerkeley conference.
In this TEDx talk, Art Instructor Steven Zapata discusses the potential ethical concerns surrounding the use of AI-generated art. One major concern is the way that AI models are trained off of example images, which are often creative content scraped from the web without consent or compensation given to the creators.
Listen to the audio version here:
Imagine we’re a hundred years in the future. Actually no. The way things are going, let’s talk about next year. You go through a difficult life event, maybe the end of a long relationship or a pet dies or a parent dies.
Soon after this event occurs, a series of obscured processes begin. You start to notice changes. You see ads to download Tinder, where before they were ads for engagement rings. The local animal shelter is suddenly in your scroll.
Odd, for sure, but to be expected so far, really. But now some new things start to happen. Your phone offers to show you some impossible images. Very realistic renderings of your dog happy in doggy heaven. But your phone knows your religion. So if you’re Catholic, your dog is happy in Catholic doggy heaven. A company tastefully, tastefully reaches out to ask if you would like to have a phone conversation with your recently deceased mother. In her voice, it will sound just like her.
And they can promise you that she will apologize for that last fight that you had. Many of you will hear these things and they will strike you as gross boundary violations. But surely there are some amongst you who think that some of this sounds genuinely soothing.
Now whatever your appetite for high tech therapy, the point that I want to make is that the technology needed to make this a reality is technology that’s already here. They’re generative AI systems. And the data needed to make this a reality is data that we already give to companies. You all know that.