Podcast: Play in new window | Download (Duration: 20:40 — 18.9MB) | Embed
AI has evolved enough now that one-shotting entire deliverables is tempting, and many people online claim to be doing just that. In this episode, Chip and Gini discuss why that’s a bad idea, and how you can train your AI tool to give you less slop-like content.
Chip is fed up with receiving one too many pieces of material with no human thought applied to them, not in preparation, not in editing, not even in the decision to hit send. His take on the broader one-shotting trend is that when the result is actually good, it’s rarely a true one-shot. The person usually spent months feeding their AI tool context first, the same way a good off-the-cuff article draws on years of accumulated expertise.
Gini suggests treating AI like a capable intern. You wouldn’t send an intern’s first draft straight to a client, so don’t do that with AI either. She describes going back and forth with her tools so much that Claude once asked her to batch her edits instead of making them one at a time. Chip pushes back on the intern comparison, arguing that with enough context and consistent feedback, AI has moved closer to a capable mid-level colleague. That feedback loop is also the focus of Gini’s internal training program, Slack sessions where the team is currently learning to push back, reject, and refine AI output instead of just accepting it.
However, don’t make the mistake of thinking that because it needs your guidance, AI isn’t worth using. Being the human in the loop doesn’t make you the savior of your business. Owners who rely solely on humans without using AI are making as grave of a mistake as people who think they can replace themselves entirely with AI. [read the transcript]
Related Posts
Leave a Reply