AI that grasps language could make the Web a nicer place.
Facebook has launched a massive project to tie together several threads of conversational machine learning. The hope: huge datasets and open software will help researchers build more articulate AIs that better understand language. Ultimately that will create more convincing chatbots. Indeed, as we report today, software with a stronger grip on language could be put to use fighting harassment online, by not only seeking out problems but also casually chatting with victims.
Get The Download! Sign up here to have it delivered free to your inbox.
As ransomware dust settles, it’s time for big questions.
The WannaCry hack that crippled 200,000 computers is abating, so let's trade panic for reflection. First, who did it? Early signs point to North Korea. Second, who’s to blame for it being possible? We wonder whether the government should stop hoarding software bugs, but debate rages over whether Microsoft and affected parties should shoulder responsibility. Third, how much did it cost? Hackers scored $55,000, but analysis suggests downtime costs could reach $4 billion.
When robots take jobs, small towns may feel it most.
That’s according to a study by MIT’s Iyad Rahwan, who explains to New Scientistthat many of the easiest jobs to automate, such as checkout assistants, are proportional in number to population. Meanwhile, those that are harder to roboticize, such as brain surgeons, cluster in cities. There are outliers: Las Vegas is big but its gambling industry could be automated. Regardless, the finding reinforces the fact that we must work out how to ensure everyone benefits from automation.
"Whenever I look at an appliance, I think: what could be done to it that causes maximum damage and embarrassment?"
— Scott Borg, director of the U.S. Cyber Consequences Unit, argues that hardware engineers have a growing responsibility to design secure devices in the future.