The robot uprising just got weirder
Attendees interact with the CloudMinds Inc. Pepper humanoid robot powered by artificial intelligence (AI) at the Sprint Corp. booth during the Mobile World Congress Americas event in Los Angeles, California, U.S., on Friday, Sept. 14, 2018.
Who would win in a test of wits: a baby or an artificial intelligence? You might think you know the answer to this; certainly, decades’ worth of science fiction stories about sinister robots have indicated that the machines would have an advantage. But a recent discovery suggests that we might have been going about this all wrong — and it might be time to start betting on the babies.
That’s the big takeaway from a new report at Gothamist, which recounts the findings of a recent NYU study. That study, titled “Commonsense psychology in human infants and machines,” pitted 11-month-old babies against what its authors dubbed “state-of-the-art learning-driven neural-network models.” What both groups were tested on was their sense of why people were doing certain things.
As the article explains, the 11-month-old babies were able to recognize goals and routines while observing a series of videos — and showed surprise when those changed. The artificial intelligence, by comparison, had a harder time recognizing the same things.
The authors of the study view this as the first chapter in a much longer line of inquiry. In the paper’s abstract, they write that this work “takes the first step in testing whether human knowledge and human-like artificial intelligence can be built from the foundations cognitive and developmental theories postulate.” For those concerned about a potential robot uprising, this is encouraging news.
Thanks for reading InsideHook. Sign up for our daily newsletter and be in the know.
