AI After Productivity
When curiosity matters more than efficiency.
I retired just around the time AI was getting mainstream buy-in, with employers expecting everyone to be using it in some way to increase efficiency and productivity. In my group, a few super-users had found ways to make better (and faster) presentations, analyses, etc., but we didn’t have widespread expected use. I’m hearing that has changed in the year-plus that I’ve been gone. I have to say, I’m not sad to have missed that.
That’s not to say I was against tools that could have helped make me more efficient and competitive. But I did have a resistance to my work being done behind the scenes without a good understanding of exactly how it was happening. I like to know how things work. And I didn’t like the idea of automated work being presented as my own.
For personal use, though, I was excited to experiment with what AI can do, and I started using it in 2024, about a year before I retired. But as I look back at my earliest chats in ChatGPT, I see that most were turbo-charged web searches or travel plan research.
The larger debate over whether AI will free us from drudgery or be our downfall, while interesting in the macro, doesn’t really affect me at this point in my life. The risks are different when you’re outside the corporate world. I don’t need to be more productive. I don’t have a job to lose. But there are still downsides to AI that are impactful to a retiree. I have a vested interest in preventing (or at least slowing) cognitive decline. Any tool that promises to think for me falls into “slippery slope” territory. At the same time, I do want to make use of any tool available to make my life better – and better doesn’t mean “most productive”.
With that in mind, I've settled into a handful of uses where AI expands my thinking rather than replaces it. I don't ask it to write these journal posts or tell me what to think about. If those ideas didn't come from my own curiosity, what would be the point?
In short, I don't use AI to outsource my thinking.
AI has gradually earned four jobs in my life: 1) ongoing searches and monitoring; 2) deep dives into subjects I want to understand better; 3) reviewing legal documents; and 4) editing and challenging my assumptions.
The quality of AI's answers depends much less on clever wording than on knowing what you're actually trying to accomplish. That changed my approach completely. Before I ask AI anything, I try to answer one question myself:
“What does success look like?”
Then I make that part of the request.
1. For searches, repetitive or not – be precise in ALL of the parameters (success means finding what I actually want, not a dozen things that are close):
“Check weekly on Monday for the flights from Charlotte to Auckland departing October 15 (departure date is set) and returning November 11 (could extend to 12th or 13th); business class only, under $10,000, max 2 connections. If you don’t find any flights that satisfy these requirements, show me one that comes close and make suggestions for expanding the search.”
2. For background or a deep dive into a topic – require the response to include opposing views on the subject, and require sources be provided for further study (success means understanding, not agreement):
“I want to know more about the US’s history with Iran; give me an overview of the past 100 years, citing sources and including opposing viewpoints. Suggest books or articles for further study on my own.”
3. For document review – be specific about your questions and provide additional information such as worksheets (success means confirming or disproving):
“My lawyer created this draft of my updated will. And here is a worksheet showing my wishes for how my estate will pass. Does the will provide for what the worksheet says? Show me the relevant paragraphs and explain how the legal wording honors my wishes.”
4. For critical analysis of an idea or a written work – be specific in requiring AI to take a contrary or critical view (success means I’ve been challenged):
“Act as a VC hearing a pitch – why would you NOT invest in this idea?”
“Here’s a draft of my Christmas letter – it’s too long; find places where I’m being repetitive and suggest cuts; don’t tell me how it’s good. Tell me how it could be better.”
One final risk – I’ve had to put in some guardrails against task creep. For me, that’s the biggest danger: spending too much time with it. Having the time to explore ideas (travel, finance, decorating, politics) makes it easy to fall down many rabbit holes. And AI loves to keep every conversation going. And going. And going. I sometimes need a gentle reminder to get off the computer.
Taken as a whole, AI has been a net positive for me. Context matters. Boundaries matter even more. I'm happy to let it remove the friction of repetitive searches or deciphering legal language.
I'm much less interested in letting it remove the friction of thinking.
That still belongs to me.