I asked Claude AI to source a summary of Gianni Giacommeli’s article “What to learn in the age of AI” to suggest four “pillars of content” for me to write about consistently for the next 1-2 years.
Here’s what it suggested regarding the first pillar:
1. Augmented Thinking in Practice The article’s central concept is rich enough to sustain a long-running series. You could write about how to direct cognitive attention, frame better questions for AI, and critique AI outputs — with real examples from different industries and roles. This is the “muscle” readers need to build, and there’s no shortage of angles: executive decision-making, frontline management, creative work, etc.
So today, I’ll write about my journey researching and applying the skill of directing cognitive attention.
As a preliminary step, I asked Claude AI to unpack what it means by writing about how to direct cognitive attention.
Here’s part of what it said:
…in an AI-augmented world, one of the most distinctly human roles is deciding where to point your focus and your team’s focus in the first place. Machines can process and generate, but they still largely work on what you aim them at. So the quality of your attention—what you notice, what you prioritize, what you choose to investigate—becomes a lever of enormous leverage.
“Directing cognitive attention” as a content theme means writing about things like:
- What deserves human scrutiny vs. what can be delegated to AI — not every decision or output warrants deep human engagement, but knowing which ones do is itself a skill
- How to ask better questions — both of AI tools and of yourself, since the framing of a question largely determines the quality of the answer
- Filtering signal from noise — the article mentions “continuous filtering of irrelevant information” as something humans still do better; you could explore how to build that habit
- Recognizing when a problem is more complex than it appears — the article distinguishes between complicated problems (which machines handle well) and complex ones (which require human judgment); training yourself to spot the difference is a form of attention management
- Pattern recognition in your domain — noticing things in your environment that others miss, and writing about how that intuition is developed and preserved alongside AI use
In short, it’s about the meta-skill of deciding what matters — before you ever open an AI tool. That’s what makes it a rich pillar: it sits upstream of everything else.
Today, I’ll start with what deserves human scrutiny vs. what can be delegated to AI.
While I agree that “not every decision or output warrants deep human engagement” (emphasis added), I’d argue that every decision or output warrants human-in-the-loop interaction of some sort. That’s why I read every output any AI algorithm produces for me. It’s tedious and even at times unnecessary, but I consider it part of responsible AI use.
So, what deserves human scrutiny, and what can be delegated to AI? Giacomelli covers these questions a bit in the article. That means I get to go a bit in depth regarding what I’ve so far been able to delegate to AI.
Collaboration. I used AI to find and partner with people for personal and professional projects. I also collaborated with AI to devise strategies and tactics for business function needs. Soon, I expect AI agents will be collaborating on behalf of humans in ultra IoT fashion.
Communication. I used AI to draft and send messages to people. I also communicate with AI using apps like Wispr flow. Full-on conversations might be the next phase of vibe-coding apps and natural-world products into existence.
Creating, devising, and simulating change management plans. So far, this remains the category of task I delegate to AI the most. I use AI to create and devise long-term, short-term, and real-time/dynamic plans. And I definitely use AI for hypothetical and scenario planning. Sure it’s useful for predictive modeling—but I wonder about use cases for exploring outliers.
Creative thinking. This one gets tricky. I’ve used AI to compile various types of ideas (e.g., business ideas, writing ideas, vacation ideas, etc.). However, I don’t outsource this skill to AI as much as I used to even as recently as last year. My increase in productivity due to AI use has freed me to explore creative effort in ways that I’d prefer to retain as a human-based skill for the rest of my days.
Finding solutions for well-defined problems. This one’s the most obvious. Calculations. Dates. Logic. Surprisingly, I still find fault with AI in its detail-orientation abilities. Fact-checking, for example, remains a hurdle. Regardless, for my limited needs, speed trumps accuracy as long as I know the error rates and can account for them via human nuance.
There are several other categories, and I’m sure I’ll get to write about them in the future.