Issue 58 |š”6 Rules for Mastering AI
We need human algorithms to know when and how to use the machine ones. Here are six of mine.
IoNTELLIGENCE delivers neuroscience-backed insights to help high-performers optimize their minds, harness AI, and thrive in a fast-changing worldāin five minutes a week.
Upgrade Your Brain. Master AI. Be Future-Ready.
š§ Time Investment: 5 Minutes
š Goal: Help you leverage AI without losing your essential abilities
š Topics: Personal AI Strategies | Be Future-Ready
How many phone numbers can you remember?
I can recall around fiveāat least two of which are my own and the one that rang at the home I grew up in. This is what happens when we outsource our memories to our mobile phones. No one planned to forget those numbers; it just happened because we relied on the machine to dial them for us.
Itās an instructive analogy for the risks we run when using other mind-extending technologies. Itās especially relevant with the arrival of ubiquitous, omniscient AI.
Itās no longer a question of whether AI will impact our work and lives but how deeply and quickly. We need to learn how to use these tools strategically now, which is why I like to say that AI requires IA (Intelligent Augmentation). Those initials could also stand for āIntentional Application.ā
I tracked my use of these cognitive instruments for a week and then reflected on how to collaborate with them intelligently and intentionally. When employed correctly, AI can give you new abilities, sharpen your thinking, and free you from drudge work. However, you must guard against de-skilling, dumbing yourself down, and developing AI dependency.
How can we strike the right balance between being force-multiplied by AI and not becoming reliant on it?
Taking inspiration from that canonical self-help manual from Stephen Covey, āThe 7 Habits of Highly Effective Peopleā (āBegin with the end in mindā), here are my ā6 Rules for Mastering AI.ā
Photo by KATRIN BOLOVTSOVA
šØ The 6 Rules for Mastering AI in 2025
1. Always ask AI - but not at the beginning of anything.
Concept: AI is an always-available idea generator, editor, and business strategist just one mouse click away. Thatās its promise, but also the root of the problem. While itās tempting to turn to it at the start of every project or blank sheet of paper, thatās precisely the wrong time to do so.
AI will atrophy your thinking skills if you let it. This is particularly true when fleshing out or communicating ideas. Itās always easier to edit than to write, but there is no thinking without writing. Wrestling with concepts is how we come to understand them.
Thatās why we have to keep the hard parts. Donāt reach for AI at the beginning of anything; save it for the middle so your brain can benefit from the struggle.
The Idea in Action: I have AI evaluate my ideas rather than generate them for me. For example, I struggled to find an opening line for this essay and was tempted to ask Chat GPT for suggestions. But wanting to preserve my ability to come up with (hopefully) interesting analogies, I persevered and came up with the phone number example all on my own (We have to take our victories where we can)!
With that in mind, here are other examples of when I use AI:
Ion: Writing | AI: Proofreading my writing.
Ion: Thinking of an analogy | AI: Evaluating my analogy.
Ion: Coming up with ideas | AI: Coming up with more ideas.
Ion: Thinking of the perfect word to use | AI: Turning to AI if stuck only after trying and failing.
2. Is it a pleasure or an endeavor? Is the purpose to extract or interact?
Concept: Reflecting on an activity's ultimate goal can help you decide whether to use an AI tool. If you will benefit from the journey, consider not using AI. If all you want is āthe answerā right now, then Chat GPT it.
The Idea in Action: I love listening to podcasts. However, there arenāt enough hours in the day to digest all the ones Iāve saved (and who has time to listen to a three-hour episode?). When aiming to enjoy the experience, I download one and dive in. When I want to extract the highlights from a pod that Iāll never get around to consuming, I have an AI app that can do that for me.
The difference is significant, and not just in terms of time savings. Listening to a podcast or reading a book is an intellectual exercise that forces you to interact with the content and author. Reading a summary of an episode or book will give you the key points but without the benefit of the entire argument. Donāt stifle your curiosity just for the sake of efficiency.
3. Think Co-Pilot, not Auto-Pilot. Only automate what you hate.
Concept: I recommend focusing on augmentation over automation. That said, there are some specific instances where outsourcing an entire activity to AI might be warranted. It should fulfill two criteria: doing it yourself adds little value (intellectually or otherwise), and you actively dislike it.
The Idea in Action: Start with your most annoying, repetitive tasks. What would you love to delegate in your life and work? Then, ask which of those activities are big pain points to remove or skills you donāt want to retain. For me, one would be the laborious task of converting my newsletter draft to the Substack format. If I could train a GPT agent to do that for me, Iād do it in a heartbeat.
4. Treat AI like an advisor, not an authority.
Concept: Donāt defer to AIās suggestions every time or even most of the time. AI can provide a second set of āeyesā on something youāre working on or serve as a tireless thought partner. However, you are the final arbiter of whether its advice is worth keeping. Itās not just a question of exercising judgment, a critical human superpower we must preserve. We also want to retain our agency, and getting in the habit of saying ānoā is a reflex we have to build intentionally.
The Idea in Action: I regularly override Grammarlyās recommendations when it strips my personality from the prose. Case in point: It wanted to ācorrectā that last sentence, but I ignored it!
5. Own, donāt outsource, the last mile.
Concept: Gordon Ramsay supervises a whole team of sous-chefs at his restaurants, but he reviews every plate before it goes to the floor and checks the finishing touches. Thatās what we need to do with AI.
The Idea in Action: When I have AI proofread email drafts, I often like some of its suggestions. However, I never copy-paste and press send; instead, I go line by line to decide which part of my version to keep and where AI did it better. Then, I read the whole thing to ensure it still sounds like one voice.Ā
6. Be monogamous at first, then polyamorous.
Concept: At the start of your AI life, you should only have eyes for one partner. Once you feel comfortable using that one AI tool, walk on the wild side and start a (working) relationship with another.
Jeremy Caplan of the excellent Wonder Tools Substack has a helpful framework for this initial, monogamous phase:
Pick: Select one tool at a time to try based on a specific need you have.
Stick: Use it consistently for a set period to thoroughly evaluate its effectiveness with real-world tasks.
Dig: Explore the tool's capabilities beyond its basic features. Experiment with different uses and applications to learn its strengths.
The Idea in Action: I recommend Chat GPT as your first model. After picking, sticking, and digging that one, try out Claude or Perplexity. Getting comfortable using two models will help you discover their respective superpowers, and itās a way of becoming ambidexterous with AI.
šØ Key Takeaways
If you found this post interesting or valuable, please share it with friends or click theĀ ā¤ļø button so more people can discover it.Ā Thanks!
IoNTELLIGENCE by Ion Valis. I'm aĀ strategic advisorĀ andĀ performance coachĀ toĀ entrepreneurs and executives. To learn more about my work, visit my website and connect with me on LinkedIn.