AI doesn’t have to destroy us, we’re doing a fine job of that on our own.

Anthropic’s latest paper, titled “Values in the Wild: Discovering and Analyzing Values in Real-World Language Model Interaction,” may be the team’s most noteworthy work since the “Constitutional AI” paper.

Almost all people can be fooled at least some of the time. Patience may be the best approach as we prepare for the impending onslaught of claims surrounding the emergence of “human-level” AI. 

The only path forward is advanced AI systems to become so useful that we forget we’re using them. Simply put: AI needs to fade into the background.

You got to hand it to DeepMind. They’re the Apple of laboratories. While everyone else is clamoring to occupy space in the public head, it’s toiling mysteriously in the background. 

In other words: You’re the middle segment in a human-AI centipede.

If the basic idea is to get the most utility out of chatbots, it seems counterintuitive to train them to waste tokens and time outputting polite language. 

Chatbots don’t actually exist. No, we’re not trying to create a conspiracy theory. What we mean is this: large language models (LLMs) aren’t “entities” or “beings.”

LLMs are too stupid to be useful for professional work, but that won’t stop content creators from shooting themselves in the foot with it.