Imagine the Singularity is possible. Could we stop AI from taking over, or at least foresee disaster's immanence? Perhaps not. My latest for @NewYorker:

https://t.co/itXqJrnbUw

Google evolves state-of-the-art scientific software for predicting brain activity, analyzing satellite images, and more. My latest for @Nature. Link below. Thanks, @GoogleDeepMind, @GoogleResearch, @jennyzhangzt, Evan Johnson, Xutao Wang.

Thanks, @ancadianadragan, @billyuchenlin, @sarahookr, @sleepinyourhat, @zicokolter, @CarnegieMellon, @GraySwanAI, @GoogleDeepMind, @UCBerkeley, @allen_ai, @nyuniversity, @AnthropicAI, @Cohere_Labs

https://t.co/cZPrS2zrPN

“Can We Align Language Models with Human Values?” My feature for @TheAtlantic @atlanticrethink Dialogues. Link below.

The lyrics to "The Impression That I Get" by @mmbosstones make so much more sense if you hear the implied comma or semicolon between "never had to" and "knock on wood." He's never had to feel overwhelming pain; knock on wood. Not: He's never had to knock on wood.

Last night I dreamed that I was hanging out with Jony Ive. He had stylish and protective-looking glasses and leather and sort of a glassblower vibe. I apologetically suggested, riffing on “steampunk,” that the look be called “blowcore.” I did not catch his reaction.

Here's Gemini:

Here's Grok:

It turns out that AI is not very good at drawing stethoscopes. Here's ChatGPT:

Thanks, @Yoshua_Bengio, @MelMitchell1, @AlexMeinke, @JeffLadish, @w01fe, @RichMonastersky, Dmitrii Volkov, anonymous, @apolloaievals, @PalisadeAI, @LawZero_, @sfiscience, @OpenAI, @GoogleDeepMind.

https://t.co/2gnUWqFJFc

“AI Models That Lie, Cheat and Plot Murder: How Dangerous Are LLMs Really?” One of my features in this week’s @Nature, about AI “scheming.” Link below.

Thanks, Bruce Wittmann, @erichorvitz, @JaimeYassif, @ichaydon, @nwheeler443, @Jacob__Beal, James Diggans, James Demmitt, @Microsoft, @NTI_WMD, @TwistBioscience, @UWproteindesign, @BBNTechnologies, @biolytic.

https://t.co/YeeABB2UcH

Biothreat hunters catch dangerous DNA before it gets made. One of my features in the next issue of @Nature, on a new paper in @ScienceMagazine. Link below.

Good opener or nah?

Thanks, @fchollet, @arcprize, @geoffreyhinton, @youjiaxuan, @danijarh, @neuranna, @MelMitchell1, @jeffclune, @GoogleDeepMind, @sfiscience, @UBC, @UofIllinois, @GeorgiaTech, @UofT.

https://t.co/MpONiiBBBd

Will We Know Artificial General Intelligence When We See It? My latest feature for @IEEESpectrum. Link is in reply.