‘We must slow the pace’: CEO of Anthropic calls for an AI slowdown

The script

Panel 1

INT. CONFERENCE ROOM - DAY Alex, his hands clasped in excitement, is presenting a new AI system to the room. Sam, his arms crossed, leans back with a skeptical look.

Alex: Imagine, Sam! We can create an AI that not only solves complex problems but also engages in thoughtful conversations!

Sam: Thoughtful conversations? What does that even mean? More chatbots, Alex?

Alex: Wait, let me finish! I'm talking about an AI that can analyze human emotions and respond with empathy. It's like having a personal therapist in every room!

Sam: Uh-huh. So it’s just better customer service, then?

Panel 2

INT. CONFERENCE ROOM - DAY Alex is now animatedly explaining his third point of his three-part plan.

Alex: The third part? Well, we'll have third-party evaluations to make sure no AI system crosses the line of harmful behavior!

Sam: Evaluations? Like how do we actually do that? There’s no way to be 100% sure. Plus, it would take a lot of time and resources.

Alex: We'll use AI to evaluate AI. Isn’t that smart? It’s like having a super-smart AI judge!

Panel 3

INT. CONFERENCE ROOM - DAY Robin, the chaos catalyst, bursts into the room with a mischievous grin.

Robin: What you’re proposing? It’s like the AI version of ‘We must slow down’ but with more speed! Let's add a little chaos to the system to make sure it really thinks before it acts!

Alex: I... I see what you did there, Robin. You’re right. Let’s add a little chaos to make sure these AIs think twice about their actions. After all, we need them to be unpredictable sometimes to keep everyone on their toes!

Sam: You’re a genius, Alex. Let’s implement the third party evaluations and the added chaos. It’s like we’re trying to create a balanced AI system—just not one that’s too calm and orderly.