TRUST BUT VERIFY [especially agentic workflows]. What an #ILTACON! The biggest takeaway is how LLM-backed tools can *help* human processes, not replace them. Even agentic workflows (especially agentic workflows) need human validation. Machines + Humans = Better than each separately.
Damien: I use AI to draft, to create, to brainstorm and to review (and sometimes revise) what I've written. I never trust it, however, to do research or to do anything that I can't double-check and correct if necessary. It's a reasonably competent assistant, but not a lot more than that, yet.
Agentic workflows heighten the need for human oversight and review. Having said that, how can humans feel confident in reviewing the AI decision points without giving up the productivity gained? Is this the 'last-mile' problem or the 'every-mile' problem?
These should have been in the giveaway bag!
Great to catch up there!
❤️Machines + Humans = Better than each separately.
Lawyer + Speaker + Writer + Builder + Mediocre Coder + Musician + VP Solutions Champion
4moAnd thanks, R. Amani Smathers, for the excellent Swiftie — really a Glassmeyer "Glassie" — bracelet!