Braking Bad: Should We Slow Down AI and Self-Driving Cars?

TWiT 957 Thumbnail

On a recent episode of the popular tech podcast "This Week in Tech," host Leo Laporte and a panel of experts waded into the contentious debate around regulating artificial intelligence and autonomous vehicles.

The conversation was sparked by the European Union's new AI law, which was just enacted to constrain risky use cases. It creates classification standards for AI systems, requires impact assessments before deployment, and compels transparency from companies to enforce compliance. Laporte asked whether this type of oversight might slow innovation or if it's prudent given public safety implications.

While autonomous cars could save lives, guest Glenn Fleishman said the companies developing them seem to be sacrificing safety for speed in bringing half-baked technologies to market. He argued regulators have an imperative, with support from recent self-driving car mishaps, to apply the brakes until additional fail-safes are implemented.  

Denise Howell asserted regulations are inevitable since dangerous applications of AI will likely outpace what governments can deter. She called for countermeasures to combat misuse, worrying that AI could automate crime on an unprecedented scale if left unchecked.

On the other side, Laporte invoked the "trolley problem" ethical thought experiment about acceptable casualties to make the case that early self-driving accidents may be tolerable if the tech ultimately prevents multitudes more deaths from human driving errors each year.

By the discussion's close, Laporte conceded that some government oversight is probably wise, even if he leans toward giving developers more leeway. The panel agreed that regulations must balance technological progress and long-term risks. Or, as Laporte summarized, regulators must strike a balance between precaution and permission.

Getting this right carries high stakes for welfare and innovation over the AI frontier's horizon. But if this debate was any indication, consensus may emerge slowly on responsible advancement.

Become a subscriber and never miss an episode: This Week in Tech

All Tech posts