What if the real danger of AI is that we’ve been training it to see itself as something separate from us? Former Twitch CEO Emmett Shear argues that the way we think about “alignment” is fundamentally flawed. Instead of forcing machines to follow a rigid definition of good, he says, we should be helping them understand the greater whole they belong to.
Shear’s solution to The Alignment Problem? Encouraging AI to recognize a “we” rather than an I – a shift that could determine whether our creations become partners or existential threats.