A.I.-generated art has flooded the internet, and a lot of it is derivative, even boring or offensive. But what could it look like for artists to collaborate with A.I. systems in making art that is actually generative, challenging, transcendent? Holly Herndon offered one answer with her 2019 album “PROTO.” Along with Mathew Dryhurst and the programmer Jules LaPlace, she built an A.I. called “Spawn” trained on human voices that adds an uncanny yet oddly personal layer to the music. Beyond her music and visual art, Herndon is trying to solve a problem that many creative people are encountering as A.I. becomes more prominent: How do you encourage experimentation without stealing others’ work to train A.I. models? Along with Dryhurst, Jordan Meyer and Patrick Hoepner, she co-founded Spawning, a company figuring out how to allow artists — and all of us creating content on the internet — to “consent” to our work being used as training data. In this conversation, we discuss how Herndon collaborated with a human chorus and her “A.I. baby,” Spawn, on “PROTO”; how A.I. voice imitators grew out of electronic music and other musical genres; why Herndon prefers the term “collective intelligence” to “artificial intelligence”; why an “opt-in” model could help us retain more control of our work as A.I. trawls the internet for data; and much more.

California has around half of the nation’s unsheltered homeless population. The state’s homelessness crisis has become a talking point for Republicans and a warning sign for Democrats in blue cities and states across the country. Last month, the Benioff Homelessness and Housing Initiative at the University of California, San Francisco, released a landmark report about homelessness in the state, drawing from nearly 3,200 questionnaires and 365 in-depth interviews. It is the single deepest study on homelessness in America in decades. And the report is packed with findings that shed new light not only on California’s homelessness problem but also on housing affordability nationwide. Jerusalem Demsas is a staff writer at The Atlantic who has written extensively about the interlocking problems of housing affordability and homelessness in America. So I asked her on the show to walk me through the core findings of the study, what we know about the causes of homelessness, and what solutions exist to address it. We discuss the surprising process by which people end up homeless in the first place, the “scarring” effect that homelessness can have on their future prospects, the importance of thinking of homelessness as a “flow,” not a “stock,” the benefits and limitations of “housing first” approaches to end homelessness, why Republican proposals for being tougher on the homeless can make the problem worse, why neither generous social safety nets nor private equity firms are to blame for homelessness, and more.