Director’s Letter June 2023
Will artificial intelligence save or destroy the world?
With the GPT-3 chatbot writing college essays and computer programs for students desperate to reduce homework loads, concerns about artificial intelligence (AI) destroying humankind have taken on a new urgency.
One version of AI-driven doomsday involves a computer program deciding on its own to eliminate humans. This ascribes an intentionality to AI that seems unlikely. Hundreds of millions of years of evolution have not only generated organisms capable of predicting and navigating the world, but organisms driven to reproduce. Lacking such evolutionary pressure, AI seems unlikely to develop a “selfishness” revealing humans as a problem to be solved through elimination.
More narrowly, as Stephen Marche noted in his essay “Of God and Machines” (Atlantic Monthly, Sept. 15, 2022), AI does what you literally tell it to do, not what you intend it to do. Consider the apocryphal story of the Air Force conducting warfare simulations using an AI-supported drone. The drone operator introduced challenges to explore AI’s ability to dynamically respond. With the ability to complete missions autonomously, the drone eliminated the operator, eliminating the emergence of new barriers. If we naively ask AI to save the Earth, would it accomplish the mission by eliminating humans?
But perhaps AI will save the world for humans. Dr. Ian Breckheimer (see adjoining article) is harnessing machine learning (ML), to better predict the world. Integrating data streams from satellites, planes, drones, and embedded sensors, he has been predicting snowmelt date for trails and meadows across the Gunnison Basin. He is extending these predictions into the world of biology, such as when plants start and stop growing, and how bees interact with wildflowers.
Dr. Breckheimer is closing the “ecological observing gap”. With human health on the line, we have invested enormous sums in understanding what happens inside bodies, from fruit flies to humans. On the scale of football fields and up, a fleet of expensive satellites constantly surveils the world. But from food productivity to carbon flows, important things happen between human bodies and football fields. This is the scale at which field biologists often work. But with notebooks and pens, there is a limit to what field scientists can measure. Until now!
Science has returned magic to the world. Arthur C. Clarke famously noted that “Any sufficiently advanced technology is indistinguishable from magic.” Chatbots such as GPT-3 use hundreds of billions of parameters, or pieces of information, derived from training languages to mimic any type of human conversation. The programs succeed in ways that are inscrutable to us, not like magic, but as magic.
Almost 600 hundred years ago, explorers started spilling across the world. Those mysterious places where the dragons lived were slowly erased from our maps. However, with the emergence of AI, the mystery, and magic, has returned, with enormous potential to do good!
Ian Billick | PhD
Executive Director, RMBL