A protester holds a 'Don't let AI take our future' placard during the demonstration by the group Pause AI outside Downing Street. (Photo by Vuk Valcic / SOPA Images via AP)When Meta agreed to a $17 billion settlement in its historic social media addiction trial last month, I was excited to see one particular detail in its concession: Facebook and Instagram will stop showing minors their number of “likes.” Why do I care about the “like” button? Because I helped create it.As a young, optimistic engineer at Facebook, I hoped likes would be a way to connect with friends. But there’s a reason this well-intentioned feature went so wrong that it became a central piece of evidence in a child safety trial: economic misalignment. Addiction is more profitable than connection, so the button became a tool to capture our attention, drive unhealthy comparisons, and keep us obsessively refreshing our feeds for tiny hits of dopamine.Since those early days, my colleagues and I have warned of the damage profit-maximizing social media platforms cause. Now, I see a misaligned technology that’s becoming an even greater threat: AI.When people in my field discuss AI alignment, they generally focus on technical alignment: the disconnect between what we want, what we ask for, and what we actually get. Imagine asking for a clean house and getting a home flooded to the rafters with bleach.Technical alignment is critical. But even if AI does what we ask, what will we ask it to do? Whatever the economy…

Because you read this