In an event that might best be described as humans versus helpful robot overlords, the United States Congress decided it was time to summon the heads of leading artificial intelligence companies to Capitol Hill for a friendly grilling that felt less like a collaboration and more like a deeply skeptical parent asking to see the browser history.
Leading executives, including OpenAI CEO Sam Altman and Anthropic’s Dario Amodei, found themselves facing the House Oversight Committee, where lawmakers expressed what could diplomatically be called “concern” over where exactly all this AI innovation might be headed. The mood in the room was part cautionary tale, part technological anxiety dream, as committee members took turns alternating between futuristic questions about synthetic biology and the more immediate issue of whether your toaster might start writing sonnets about its existential dread.
OpenAI’s Altman, who once predicted that artificial intelligence could be humanity’s most transformative invention, struck his usual tone of measured optimism but was met with bipartisan furrowed brows. Fears ranged from AI-generated misinformation knocking elections sideways to deepfakes upending reality with the ease of a Photoshop filter. Somewhere in there, someone likely wondered aloud whether AI could be stopped from writing vaguely existential haikus about tax reform.
The hearing featured a few predictable calls for tighter regulation, with some lawmakers suggesting that AI companies should be treated less like plucky startups and more like chemical factories or nuclear power plants, which is always a fun comparison when you are just trying to deploy a chatbot that can tell you why your sourdough starter keeps dying.
“We are in a race against time,” declared one committee member, possibly picturing an AI robot sprinting in loafers toward the 2024 election.
The executives nodded solemnly and mostly agreed that yes, some degree of oversight might be in order, though none seemed too eager to hand Congress the metaphorical keys to the algorithm. Instead, they politely suggested that the best approach might be a blend of private self-regulation, public policy and a generous helping of hope.
In a final turn that surprised exactly no one, lawmakers ended the session by acknowledging that they themselves might need a bit of AI help just to understand what it is they are trying to regulate. One can only look forward to the inevitable moment when Congress asks GPT-4 to write the legislation meant to restrict GPT-4.
And so Capitol Hill’s latest episode of Humans Being Slightly Alarmed continues its run, now in ultra-high-definition machine learning clarity.