General Intelligence
Google’s Voice A.I. Can Tell Apart People Talking Over Each Other
And other fascinating A.I. research from the week
The Pentagon announced A.I. ethics principles this week, meant as an exercise in responsible thinking about the impact of artificial intelligence in war.
The principles are underwhelming.
Perhaps the strongest language in the five principles is a call to “exercise appropriate levels of judgment and care,” which we can all agree means very little. Appropriate to whom?
Other principles outline the absolute minimum we could ask for from an algorithm that might be deployed to take human life, like our ability to shut them off if they’re acting in unintended ways.
These are not A.I. principles as much as they are justifications for the U.S. military to use however it sees fit.
“I worry that the principles are a bit of an ethics-washing project,” anthropologist Lucy Suchman told the AP. “The word ‘appropriate’ is open to a lot of interpretations.”
Given a list of ingredients, this algorithm tries to generate an image of a dish containing those ingredients.
