
Fink explains his work building fully autonomous vehicles for the DARPA Grand Challenge. He explains how massively parallel computer systems with ultra-dense vision arrays can map terrain in real time, detect obstacles by type, and anticipate the behavior of humans, animals, and other objects. Fink describes building ethical decision-making into the lowest software layers so a vehicle could override an impaired driver or minimize accident damage.
Though his team was cut from the DARPA competition despite praise for their software architecture, Fink announces the formation of the International Robot Racing Federation. He outlines plans for a September race in Nevada that would be open to international teams and serve as a proving ground for autonomous vehicle technology that could eventually reduce highway fatalities by two-thirds.
Key Moments
Sentience as Subtler Pattern Recognition: Fink frames machine sentience not as mystical consciousness but as a computer's ability to recognize esoteric, high-level patterns the way humans do - extending beyond simple object recognition into the kinds of behavioral and emotional patterns people perceive in each other.
The Forbin Project Scenario: Art poses the Forbin Project thought experiment: a defense AI concludes that humanity's behavior is so suicidal - population, ecology, atmosphere - that for our own good, under the first law of robotics, it makes decisions that are 'expedient and deadly for many.' Fink agrees that's the dilemma.
Fink's Personal Code: Asked how he'd avoid the Forbin trap, Fink says he is a great believer in free choice and human progress through choice, and that he is completely against any machine limiting that - vowing it would be 'a cold day in the hot regions' before he wrote code that would let a computer restrict a person's freedom.
Should a Robot Stop a Suicide?: Art poses a deliberately hard question: how would a robot respond to its owner committing suicide? Fink's answer, after a pause, is that the robot would have to allow it - the same free-choice principle he just laid out, taken to its uncomfortable conclusion.
Refusing to Put a Gun On It: Pushed by Art on whether autonomous battlefield supply vehicles in Iraq would have to make life-or-death targeting choices, Fink concedes DARPA may buy his technology and weaponize it, but says their stated purpose is supply lines and medicines - and he personally won't be the one to put a gun on the platform.
