“Once We Create It, It May Be Out Of Our Hands Entirely”

Marc Goodman, law-enforcement veteran and author of the forthcoming book Future Crimes, sat for an interview with Jason Dorrier of Singularity Hub about the next wave nefariousness, Internet-enabled and large-scale. A question about the potential for peril writ relatively small with Narrow AI and on a grand scale if we create Artificial General Intelligence. An excerpt::

Question:

Elon Musk, Stephen Hawking, and Bill Gates have expressed concern about artificial general intelligence. It’s a hotly debated topic. Might AI be our “final invention?” It seems even narrow AI in the wrong hands might be problematic.

Marc Goodman:

I would add Marc Goodman to that list. To be clear, I think AI, narrow AI, and the agents around us have tremendous opportunity to be incredibly useful. We’re using AI every day, whether it’s in our GPS devices, in our Netflix recommendations, what we see on our Facebook status updates and streams—all of that is controlled via AI.

With regard to AGI, however, I put myself firmly in the camp of concern.

Historically, whatever the tool has been, people have tried to use it for their own power. Of course, typically, that doesn’t mean that the tool itself is bad. Fire wasn’t bad. It could cook your meals and keep you warm at night. It comes down to how we use it. But AGI is different. The challenge with AGI is that once we create it, it may be out of our hands entirely, and that could certainly make it our “final invention.”

I’ll also point out that there are concerns about narrow AI too.

We’ve seen examples of criminals using narrow AI in some fascinating ways. In one case, a University of Florida student was accused of killing his college roommate for dating his girlfriend. Now, this 18-year-old freshman had a conundrum. What does he do with the dead body before him? Well, he had never murdered anybody before, and he had no idea how to dispose of the body. So, he asked Siri. The answers Siri returned? Mine, swamp, and open field, among others.

So, Siri answered his question. This 18-year-old kid unknowingly used narrow AI as an accomplice after the fact in his homicide. We’ll see many more examples of this moving forward. In the book, I say we’re leaving the world of Bonnie and Clyde and joining the world of Siri and Clyde.•

Tags: ,