The News Tonight is AI Danger

Top story: CNN, New York Times, Washington Post, BBC, CBS, NPR.

Visible without scrolling: Wall Street Journal, Forbes.

Scrolling to 2nd-5th screen: MS Now, Fox News, CNBC, Yahoo! News, Reuters.

Not on first 5 screens: New York Post.

This is not my filter bubble! I started with a pretty clean Chromium browser, deleted all browsing data again, then visited all the mainstream news sites I could think of. (I otherwise use a pretty heavily-customized Firefox.)

This feels similar to abrupt change in news coverage from early March 2020. Before then, I was one of a few nerds following a niche series of concerning events unfolding over the past few months. Then we had a nucleation point and it’s suddenly center stage.

Yesterday, my mom asked me what I thought of Bernie Sanders’ warning on AI. I don’t generally follow him, but looked it up, and had to say yes, it’s all at least directionally correct. (My only nitpick was that it seems pointless to request that the major AI labs pause training and deploying new models. The labs are begging for regulation that would require them to slow or stop, but seem incapable or unwilling to pause without that.)

Among other feelings is relief. Centers of power are perhaps taking this as seriously as I’ve felt was appropriate since watching that Black Hat talk with the misaligned agent swarm on August 7, followed by a drumbeat of other incidents seemingly every few days, most recently yesterday.

Hard problems remain unsolved! But one of them, expanding the overton window to talking frankly about misaligned AI danger with non-technologists, might be kinda solved now.