Hi Subscriber! Thanks for being here. Here is a helpful link to the entire Front of the Check Archive. Go back and see what’s useful. It’s all yours. https://www.frontofthecheck.com/archive
“Assume this is the worst AI you will ever use.” Ethan Mollick
Featured Story
The Signature Stack. Four Steps for Thinking with AI
A New Line of Thinking Is Required for Today's AI Work
Sherman Mohr | INSEAD Blue Ocean Strategy Certified Practitioner and Over 50 Pro
I’m intentionally leaving this here for another week.
AARP's 2026 research found that eight in ten adults over fifty have used some form of AI, yet most rate themselves beginners or intermediates, and nearly four in ten don't trust it to deliver accurate information.
What I’m witnessing is the typical experienced professional claims they use the tools. They distrust it a little, and have a mindset that the gap between them and the 28-year-old down the hall is a gap in technical skill. Many think the solution is more prompts, better prompts, a weekend course with "engineering" in the title.
No, no, no.
Prompts are syntax. Syntax expires. David Bullock and Sue Bell, in How to Think with AI, make this point: building your competence around today's tools is short-sighted, because those tools will be obsolete. Wharton's Ethan Mollick goes further in Co-Intelligence with a rule worth taping to your monitor: assume this is the worst AI you will ever use. If the model in your hand is the worst it will ever be, a skill tuned to its quirks is a depreciating asset. You are memorizing the menu at a restaurant that reprints it every quarter.
What doesn't depreciate is how you think. Specifically, how you think alongside a machine that is fast, fluent, tireless, and occasionally, cheerfully, wrong. That requires a new line of thinking. Not a new tool. A new operating model.
In June, we argued that judgment is the asset AI can't take away from you. This is the sequel, and it carries an uncomfortable corollary: judgment you can't access under pressure is worth exactly as much as judgment you don't have.
The evidence: AI makes calm experts better and stressed operators worse
Read three studies together and a pattern emerges that no single study states outright.
The jagged frontier. Harvard Business School and Boston Consulting Group ran one of the largest controlled experiments on AI at work: 758 BCG consultants, randomly assigned to work with or without GPT-4. On tasks inside the model's capabilities, the AI group completed 12.2% more tasks, worked about 25% faster, and produced work rated roughly 40% higher in quality. Then the researchers handed them a task that looked similar but sat outside what the AI could do well. The AI-assisted consultants were 19 percentage points less likely to get the right answer than colleagues working alone. The machine didn't just fail. It talked smart people out of their own judgment.
The confidence trade. A Microsoft Research and Carnegie Mellon study of 319 knowledge workers found an inverse relationship that should stop every reader in their tracks. The paper's own summary: "Confidence in AI is associated with reduced critical thinking effort," while confidence in oneself is associated with more of it. Workers who lacked the skills to inspect, improve, and guide AI output tended to skip the critical thinking altogether.
The pressure tax. You've seen this one here before. In our September piece on starting over after financial reversal, we cited the Princeton-Harvard research published in Science: financial pressure alone cost lower-income participants the cognitive equivalent of roughly 13 IQ points.
Now connect them. AI is an amplifier, and it amplifies whatever is in the operator. A calm, experienced, self-trusting professional gets leverage. A stressed, self-doubting one gets confidently wrong output at speed and loses the very confidence that would have caught the error.
That is about tech. It is about your nervous system.
For those of us over 50.
Two forces are colliding in your career right now.
The first is leveling. In the BCG study, the consultants who started weakest gained about 43% from AI; the strongest gained about 17%. AI lifts average work toward good. That sounds democratic, and it is. It also means "solid and reliable", the reputation many of us built careers on, no longer differentiates. The premium migrates to the territory outside the frontier: ambiguity, thin data, high stakes, reading people. That is the space where thirty years of pattern recognition earns its keep.
The second force is the transition itself. Layoffs, a pushed-out exit, a first consulting launch, a money squeeze, these are high-pressure states, and high-pressure states degrade judgment. Angus Nelson, author of Neuro Resilient Leader and a past Over 50 Voices subject, has a name for it: Decision Degradation, the gradual erosion of decision quality under sustained pressure. His point is that it isn't a character flaw. It's a mechanical problem with a mechanical fix.
So here is the scenario I see over and over. A capable 56-year-old, three weeks after the exit, 11:40 p.m., asking a chatbot to rewrite their LinkedIn summary for the fifth time and accepting whichever version sounds most confident. That isn't using AI. That's outsourcing your own self identity.
Nelson has a line for the hustle instinct: "You can't outwork velocity." You can't out-prompt it either.
Think of the Signature Stack as a framework.
Our masthead reads Your Life, Your Signature. That turns out to be the right frame, because the last step of thinking with AI is putting your name on something. The Signature Stack has four layers, in order. Skip one and every layer above it wobbles.
1. STATE → 2. SCOPE → 3. SPAR → 4. SIGN
Layer 1 — STATE: Regulate before you prompt
Nelson's argument in Neuro Resilient Leader is that the most expensive bottleneck in most businesses isn't strategy or systems. It's a dysregulated nervous system. When your internal operating system runs on scarcity and fear, decisions stall and the gut instinct you've relied on for decades goes quiet. He lists the tells: a clenched jaw, shallow breathing, decisions that suddenly take longer.
His C³ Protocol gives you three checkpoints:
Clarity — who you are, what you want, and who you must become.
Capacity — your ability to carry responsibility and pressure without buckling.
Composure — staying steady under load. Nelson describes it as behavioral: what you actually do when the pressure lands.
Here is the systems connection. The Microsoft study shows self-confidence drives critical thinking. Self-confidence isn't only a trait. It's a state, and the state has to be regulated. Regulation is how you get access to the expertise you already own.
The rule: Never ask a machine to decide something you're too rattled to evaluate. If you fail the C³ check, the highest-leverage move is a walk around the block, not a prompt.
Layer 2 — SCOPE: Map the frontier before you trust the output
The BCG researchers called it a jagged frontier because tasks that look equally hard to a human can sit on opposite sides of the line. You cannot see it from a distance. You find it by walking it.
That's the logic of Mollick's first rule: always invite AI to the table. Not because it will be right, but because experimentation is the only way to learn where the frontier runs through your work.
Build a personal frontier map with two columns:
Inside (for now): drafting, summarizing, reformatting, generating options, first-pass research you then verify, rehearsing a difficult conversation.
Outside (for now): reading a room, pricing a trade-off where the data is thin, sensing which client is about to walk, anything where the answer depends on context only you hold.
Re-draw it every quarter. The frontier moves, and per Mollick, it moves in one direction.
The decision filter: If this answer were wrong, would I know? If the honest answer is no, you are outside your frontier, even if you're inside the AI's.
Layer 3 — SPAR: Converse, don't command
This is where Bullock and Bell's work earns its title. How to Think with AI asks readers to abandon the hunt for the perfect one-off prompt and instead hold an ongoing, iterative dialogue: AI as a thinking partner that refines ideas and challenges assumptions. Their most underrated idea is using AI to surface and mitigate your own cognitive biases, not just the machine's. Tellingly, they structure the book as a ten-week cognitive boot camp. Thinking with AI is trained, not installed.
Bullock put it simply when we revisited him in June: the best use of AI, he said, "is not to stop thinking. It is to think more clearly." He added that people with real experience often use it better, because they already know what deserves weight.
Three sparring moves:
Put your position on the table first. Write your raw take before you open the window. If you can't, you have nothing for the machine to sharpen.
Cast the opponent. Mollick's advice is to treat AI like a person, and to tell it which person. Not "helpful assistant." Try the skeptical CFO who has to approve this, the prospect who already said no twice, or the board member who reads footnotes for sport.
Ask for your blind spots, not your applause. "What am I assuming that a smart person from a different industry wouldn't?" "What's the strongest case against this?" A model asked for validation will supply it in bulk. Validation is the cheapest thing it makes.
Layer 4 — SIGN: Own the verdict
Mollick's second rule is to be the human in the loop. The Microsoft researchers describe what that means in practice: AI shifts knowledge work from execution to oversight, verifying information, integrating responses, and stewarding the task.
Read that like a job description, because it is one. You are no longer the person who writes the first draft. You are the person who decides whether the draft is true, whether it fits, and whether you'll stand behind it. Composure, Nelson's third C, shows up here: signing means accepting the consequence.
The signature test — three questions before anything AI-assisted leaves your hands:
Can I defend every claim in it without the chat transcript open?
Does it contain at least one thing only I could know?
Would I sign it in ink?
Two out of three means it isn't finished.
The stories that are costing you. Stop saying these things.
Four stories show up in nearly every conversation I have with readers about AI. Each has a price.
"I'm not a tech person." Cost: abstention. You stay a permanent beginner in a skill that is mostly judgment, the thing you're best at.
"I'm too late." Cost: the same, plus resentment. The tools reset every few months. Nobody is early, and everybody is late.
"It's just a better Google." Cost: misuse. You ask for answers instead of arguments, and you get the frontier's worst outcome, fluent and wrong.
"It'll replace judgment anyway, so why bother." Cost: surrender. The evidence runs
What Waiting Costs
Silence has a cost, and here it compounds. Every quarter the models improve, the median worker gets better without lifting a finger, and the bar for "good enough" rises. Standing still is falling behind, not against the machine, but against the people using it well.
There is a second-order cost that's harder to see. The professional who builds the Stack now builds habits of regulation, scoping, and sparring. The one who waits builds nothing. They build habits of deference, one midnight prompt at a time. Habits compound in both directions.
HOMEWORK
Pick one consequential decision you're facing this week, pricing a first engagement, choosing which role to pursue, shaping a pitch. Run it through the Stack in a single 45-minute session:
STATE (5 minutes). Write one sentence each for Clarity, Capacity, and Composure. If one is red, reschedule the session. That's not avoidance; that's the protocol.
SCOPE (5 minutes). List which parts of this decision sit inside the frontier and which sit outside.
SPAR (25 minutes). Write your position first. Then cast the AI as your toughest skeptic and ask for your three biggest blind spots.
SIGN (10 minutes). Close the AI window. Write the final decision in your own words. If you can't, you aren't done thinking, and that's the most useful thing you'll learn all week.
Bring what you learned to Monday's Second Act Advantage Community call. Thinking with AI gets sharper in a room of peers who will tell you when the machine flattered you.
Where to start
Angus Nelson, Neuro Resilient Leader — the full digital edition is offered free at angusnelson.com/freebook.
David Bullock & Sue Bell, How to Think with AI and its companion, How to Think About AI — howtothinkwith.ai.
Ethan Mollick, Co-Intelligence — and his free newsletter, One Useful Thing.
Sources
AARP Research, "Navigating the World of AI" (data story, January 2026). https://datastories.aarp.org/2026/navigating-the-world-of-ai/
David Bullock & Sue Bell, How to Think with AI (MindShifting Media). https://www.howtothinkwith.ai/ ; companion volume How to Think About AI: Embracing the Cognitive Revolution (2024). https://www.howtothinkwith.ai/how-to-think-about-ai/
Ethan Mollick, Co-Intelligence: Living and Working with AI (Portfolio, 2024). https://www.penguinrandomhouse.com/books/741805/co-intelligence-by-ethan-mollick/
Angus Nelson, Neuro Resilient Leader: The Internal Operating System for Leading at the Speed of Change. https://angusnelson.com ; book description (Decision Degradation, C³ Protocol): https://nlb.overdrive.com/media/13200483 ; "You can't outwork velocity" — The Mindset Doctor Podcast with Dr. Justin Moseley (May 4, 2026); nervous-system tells — The Sarah Centrella Show, Ep. 116; strategy-vs-nervous-system bottleneck — Learn-It-All Podcast, Ep. 308 (July 2026). https://www.learnit.com/podcast/308-the-neuroscience-of-high-performance-3-steps-for-clarity-confidence-angus-nelson
Fabrizio Dell'Acqua et al., "Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality," Harvard Business School Working Paper 24-013 (Sept. 2023); published in Organization Science. https://www.hbs.edu/ris/Publication%20Files/24-013_d9b45b68-9e74-42d6-a1c6-c72fb70c7282.pdf
Hao-Ping Lee et al., "The Impact of Generative AI on Critical Thinking: Self-Reported Reductions in Cognitive Effort and Confidence Effects From a Survey of Knowledge Workers," Microsoft Research & Carnegie Mellon University (CHI 2025). Coverage: https://campustechnology.com/articles/2025/02/21/study-generative-ai-could-inhibit-critical-thinking.aspx
Anandi Mani, Sendhil Mullainathan, Eldar Shafir & Jiaying Zhao, "Poverty Impedes Cognitive Function," Science (2013) — as cited in Front of the Check, "Starting Over After Financial Reversal" (Sept. 22, 2026). https://www.frontofthecheck.com/p/starting-over-after-financial-reversal
Front of the Check, "You've Spent Thirty Years Building Something AI Can't Take Away," incl. Over 50 Voices revisit with David Bullock (June 30, 2026). https://www.frontofthecheck.com/p/you-ve-spent-thirty-years-building-something-ai-can-t-take-away
Join us in community? Join the Second Act Advantage Community by visiting us each Monday. No agenda, some structure, some accountability. We’ll start October 5th at 12:30 Eastern with a simple three question guideline.
1) What do you have planned this week?
2) What are your hurdles?
3) What did you accomplish over the last week?
Here is the link: Second Act Community Call
Monday, October 5 · 12:30 – 1:00pm
Time zone: America/New York
Google Meet joining info
Video call link: https://meet.google.com/xpk-xbxe-pyy
If you need assistance or want to discuss anything above, schedule a call.
Thanks to our sponsor. Nokbox: A system designed to keep life organized, even when things get chaotic. NOK stands for Next of Kin, and NOKBOX help you organize important documents, accounts, possessions, and estate details so your next of kin knows exactly what to do in something happens.
But it’s not just for the future. NOKBOX gives you peace of mind now by keeping everything in one organized place, saving time, reducing stress, and making it easy to find what you need when you need it.
Check them out below and invest in a NOKBOX for you and your family.
A system to stay consistent even when life gets chaotic.
Over 50 Pro Featured Resources of the Week:
These frameworks and diagnostic tools operationalize the shift from competing in saturated markets to designing uncontested career paths.
One Useful Thing: Wharton professor Ethan Mollick's ongoing guide to experimenting with AI. It directly reinforces the SCOPE and SPAR concepts, providing practical ways to invite AI into your workflow, map the "jagged frontier," and treat the model as a collaborator rather than a search engine.
How to Think with AI: David Bullock and Sue Bell's platform reinforces the SPAR layer of the Signature Stack. It shifts the focus away from memorizing temporary prompts and toward building an ongoing, iterative dialogue with AI to surface biases, refine ideas, and think more clearly.
Neuro Resilient Leader: Supporting the STATE layer, Angus Nelson's resource tackles "Decision Degradation." It provides the framework for the C³ Protocol (Clarity, Capacity, Composure) to help professionals regulate their nervous systems and regain access to their own expertise under pressure.
Navigating the Jagged Technological Frontier: The Harvard Business School AI Institute's summary of the landmark BCG experiment. It provides empirical evidence for why AI acts as a leveling force while simultaneously tricking smart professionals into abandoning their own judgment when operating outside the model's capabilities.
The Impact of Generative AI on Critical Thinking: Coverage of the Microsoft Research and CMU study validating the "confidence trade" and the SIGN layer. It illustrates the inverse relationship between over-trusting AI and critical thinking effort, proving that users must shift from simple execution to active oversight and verification.
Favorite Stories of the Week:
Here are four examples of professionals treating AI not as an execution machine, but as a genuine cognitive partner. These stories perfectly illustrate the SPAR (Converse, don’t command) layer of the Signature Stack:
The CEO’s "Board of Brains" (Scott Galloway): Instead of using AI to write his presentations, marketing professor and CEO Scott Galloway uses it to pressure-test them. Before presenting to his board of directors, he feeds the model his pre-read deck and prompts it to act as a hard-charging venture capitalist. He asks it for three insights and three critiques. Galloway found that the AI's feedback matched 90% of what his actual board ended up saying, allowing him to walk into the room with his blind spots already covered.
The Engineer's First-Principles Pivot (Park Junwoo): While designing a power system for space, Park Junwoo initially assumed he needed to build solar panels. Instead of asking AI to design the panels, he asked it to challenge the assumption. The AI instantly calculated the physical constraints—noting the difficulty of manufacturing high-purity silicon wafers in space—which forced Junwoo to pivot his entire architecture to solar thermal turbines made from easily accessible asteroid metals. He explicitly commands his AI: "Argue that this assumption is wrong," because, as he notes, requesting counterarguments is the only way to overcome the machine's default sycophancy.
The "Spicy" Digital Twin (GrowthPath Partners): To prevent their own thinking from getting lazy, a strategist built a custom GPT called "Digital Liza." Rather than setting it up as a helpful assistant, they gave it a specific mandate and personality: bold, salty, and designed exclusively to challenge their work and find blind spots. By forcing the AI to evaluate their ideas against strict criteria and push past default responses, they use the tool to open up their aperture to different points of view before making a final human judgment.
The Pushback Protocol (Real-world Developer): In a community discussion about using AI as a thinking partner rather than a generator, a developer shared their core operating rule for AI collaboration: they explicitly instruct the model that if they suggest a flawed idea or an inefficient approach, the AI must push back. They use a specific sparring technique: "Now give me the strongest case against the response you just gave." By forcing the model to argue against itself, they break the AI's "agreeableness loop" and surface assumptions they didn't realize they were making.
In Case You Missed It
You’re doing $15 tasks with $150 talent. Need help with spreadsheets, social media posting, article creation and posting, LinkedIn connecting, etc? How does reliable help for $5.00 hourly sound?
A free seven day challenge program that gets you reconnected with your network. Less than 10 minutes a day. Click Here
Interested in business coaching, consulting, or Fractional work? Over50Pros offers a free research report on the state of the business coaching, consulting, and Fractional Executive industries. Also, a helpful Industry Entry Decision Guide. Both resources are free with an email address. https://over50pros.com
Final Thoughts
Thought Partners
I’ve been blessed with many thought partners. Initially, my mom and dad. They helped me first but telling me what to think, how to think, and what to hold as truth. I had to grow from other thought partners contribution to my life to begin to challenge some original thinking that no longer felt congruent with whom I saw myself to be. Those thought partners were teachers, professors, and friends. One day, spouses became great thought partners. Not more than one at a time mind you. 🙂
Mentors provided thought partnership, business coaches, and consultants served their roles.
Now AI gets a seat at the table. It is a worthy thought partner. I’ve found myself reading more over the last two years. Yes, real books that you have to hold and touch. Why? I want to be an independent thinker who leverages AI as a thought partner, not a crutch.
I hope you do as well. You are worthy of great thought partners.
And as always, if you need help with anything related to topics in the newsletter or transition over 50 years old, schedule a call.
Please click below and tell me what you like or dislike about the Front of the Check newsletter and how I may deliver more value to you! You are my growing community and I want your feedback. I’ll Venmo or Zelle you $5 for two minutes of feedback.
Coming Next Week: Prepping for a presentation for your company or project.
Experience Leads to Revenue