This is a paid subscriber post, including a free summary. Free subscribers, for access to the full post, please use the button at the bottom to explore subscription options.
Starting this week, I’m going to be sharing executive takeaways from my AI work, each Friday, on LinkedIn. If you are interested, please click here to follow!
Dear Prompt Experimenters,
Agents. Task automation. A-B testing for productivity gains. Blah…
I don’t deny these opportunities are real, but I’m getting bored with them. They are all most people are talking about, and they represent an extremely limited view of what AI offers.
Let’s dig into that today. I’m going to share my own extended (yeah, this one is really long…I hope you don’t mind!) conversation with AI on this topic, and make it fully interactive, so that you can join in to agree with me, or tell me why I’m completely wrong.
We’re also going to explore AI’s “over-agreement problem.” Yes, it does tend to take your side. But we can easily change that. With the drop of a prompt, it will become your all-out intellectual adversary. It’s refreshing, and it’s useful. We are not afraid of intellectual friction here; we thrive on it!
This post will have four parts:
I. A friendly conversation about the value of AI: I began with model 4o, expressing my frustration (everyone is so fixated with AI time savings productivity, and I think this is narrow-minded), and creating a structure (efficiency vs. effectiveness) for our conversation. As expected, 4o was friendly and supportive, universally in agreement with everything I had to say. It tends to do that.
II. Getting rigorous: We pulled in model o3 pro to think about the topic more objectively and analytically, integrating hard numbers and data.
III. Intellectual battle: I asked ChatGPT to create three distinct intellectual enemies who would strongly disagree with my ideas. Not strawmen. Real, embodied, and highly competent personas, each given an opportunity to attack.
IV. Interactivity: I turned the entire conversation into a live GPT. Paid subscribers can now engage directly to ask your own questions, take sides, or extend the debate.
The conversation that follows includes adversarial AIs wrestling over the future, and a live GPT you can interrogate yourself. It’s worth the click.