--> @emojiaquarium.bsky.social

So you're saying we need one model to rule them all, one model to find them, one model to bring them all and in the darkness bind them?

Agents doing search vs. agents optimizing SEO, all fueled by marketing money. We may be building the greatest adverserial network ever.

There's an obvious role for someone who routinely checks the operation of AI agents which requires neither in-depth AI nor business knowledge, but a combination of IT knowledge and common sense. Could be that HRIT departments are coming.

True. Just as a manager isn't necessarily an expert in recruitment, personnel development or labor law, there's an obvious parallel to setting up, operating and managing the trustworthiness of AI agents.

Could be great for optimizing the logistics inside a hospital

I think part 1 is expectation management: yes, we know AI will be automating things (it has done so for quite some time now), but in practice it is more than the LLM's quality that counts and that often looks less spectacular than the hype suggests.

Part 2 is trying to keep the funding going.

Given that, according to you, 'GPT-4 (...) beat most humans at innovation,' how would you factor the risk of someone beating you to the punch into your wait calculation?