Events

Abstract: Contemporary AI systems are built by composing many pre-existing software modules that pass through many hands before becoming a finished product or service. In interviews with 27 artificial intelligence engineers across industry, open source, and academia, I found how engineers rarely see responsible AI to be within their power or responsibility to address. In this talk, I will present how fractured and distributed AI supply chains shift responsibility for AI harms to be always elsewhere, and thus nowhere. I will then present more recent work examining stops along this supply chain: how the United States Department of Defense funds research into (anything but) "Basic" AI research, and on the incentives that leave academic AI researchers to merely play Big Techs' game.