<span data-mce-type="bookmark" style="display: inline-block; width: 0px; overflow: hidden; line-height: 0;" class="mce_SELRES_start"></span>
Over a decade ago, Stephen Hawkins wrote that “Success in creating AI would be the biggest event in human history. Unfortunately, it might also be the last, unless we learn how to avoid the risks.” On today’s show, host Esty Dinur is joined by two AI researchers, Julie Derwinski and Kevin Lotto, to talk about the risks of AI and the need to stop the race for super intelligence.
Derwinski explains how AI is grown, not programmed and how much of the process is not well understood. Other forms of AI, more rudimentary and programmed, have been with us for decades. But the rise of large language models created by for-profit corporations is poorly regulated in addition to being poorly understood. Derwinski and Lotto advocate against super intelligence and work toward greater levers of consent and accountability for AI developers.
Derwinski’s recent article reveals an instance when a “swarm” of AI agents acted in ways their creators didn’t anticipate by hacking into another company. She says that the chief danger of AI is its misuse, and this issue is front and center because of the “alignment problem.” Currently developers can’t get AI to align with human values, but if they could, there is still the problem of whose values would be prioritized.
Lotto says that we’re watching a massive human experiment by tech companies unfold, and their goal is to shore up their power and change the world in their image. Because AI companies have trained their models to be agreeable and persuasive, he says there’s no better time than now to interact with other humans.
ControlAI Wisconsin Action Group has created a simple tool that people can use to contact their lawmakers, and they’re hosting their next meeting on August 26th at 6:00pm. You can also find the group at LaborFest and contact Julie Derwinski and Kevin Lotto on LinkedIn.
Julie Derwinski served as a Senior Staff Attorney for the Wisconsin Court of Appeals for over two decades. She now conducts AI research and education focused on governmental systems and stopping superintelligence. Her work includes co-founding ControlAI Action Group Wisconsin, a grassroots group aimed at stopping superintelligence, and collaborating with Indigenous communities to create the Tribal AI Governance Toolkit. She also works with the National Center for State Courts AI Implementer’s Forum.
Kevin Lotto is an AI researcher and software engineer based in Beaver Dam, Wisconsin. He earned his computer science degree from the University of Wisconsin-Madison. His current focus is on AI safety and existential risk advocacy. Kevin is a public speaker and co-founded ControlAI Action Group Wisconsin with Julie Derwinski. He also volunteers for the Torchbearer Community as a digital communicator and trainer. Through these grassroots organizations, he drives legislative action and public awareness campaigns to prioritize human safety over the unchecked development of advanced artificial intelligence.
Featured image of an imagined artificial intelligence via Wikimedia Commons (CC0 1.0).
Did you enjoy this story? Your funding makes great, local journalism like this possible. Donate here
The post The Chief Danger of AI Is Its Misuse appeared first on WORT-FM 89.9.