writing

The Convenience Trap: How AI Consumer Tools Could Be the Death of Critical Thinking — Right When We Need It Most

cantechit ai , , , , ,

Part of the ongoing AI series at cantechit.com

Let me start with something that is going to sound like a hot take but I promise it isn’t: AI doesn’t hallucinate. Not really.

I know. Bear with me.

Ever since ChatGPT blew the doors off in late 2022, “hallucination” has been the word everyone reaches for when an LLM makes something up with absolute confidence. Scholars are now pushing back on that framing — and honestly, they’re right. Cognitive scientists and AI researchers argue the more accurate word is confabulation — a psychiatric term for when a brain (or a system that mimics one) fills in narrative gaps with plausible-sounding fiction it genuinely believes is true. It’s not lying. It’s not hallucinating in the clinical sense. It’s pattern-matching on flawed or incomplete ground truth and arriving at a confident, wrong answer.

Sound familiar? It should. Because humans do this constantly.

Every time someone doubles down on a completely false “fact” they picked up from a bad source twenty years ago, that’s confabulation. Every time a witness in a courtroom sincerely misremembers what they saw — their brain filling in details that feel true — that’s confabulation. The neurons in your brain and the mathematical structures inside a large language model are working on the same fundamental principle: reasoning forward from whatever ground truth they were trained on. If the training data is garbage, the output will be wrong. Confidently, convincingly, devastatingly wrong. The AI isn’t uniquely broken. It’s human in a very uncomfortable way.

So stop calling it hallucination. Start calling it what it is. And then — more importantly — ask yourself what that means about how much you should trust either source.

The Human Context Window: Our Last Real Edge

Here’s where I want to push back on the doomers a little, because there is one genuinely huge advantage the human brain has over every AI system that exists today: our context window is enormous.

Every LLM has a context window — the amount of information it can hold “in mind” at once during a session. GPT-4 can handle about 128,000 tokens. Claude goes up to 200,000. That sounds huge until you realize your brain is running a context window that incorporates decades of lived experience, emotional memory, embodied knowledge, real-world cause-and-effect, and deeply personal relationships — all simultaneously, all the time, all connected to each other in ways that no model can currently replicate.

Yes, an AI has access to a much larger static memory — the entire training dataset — but it has to recall and retrieve from that pool every time, with no persistent working memory across sessions by default. You carry your context window with you everywhere. It never logs out.

That’s not a small thing. That’s arguably the biggest structural advantage humans have right now, and I don’t think we talk about it enough. When I’m making a judgment call at work, I’m not just running that query against a database. I’m drawing on thirty-some years of pattern recognition, failure, intuition, and relationships. No model has that. Not yet.

The danger? We’re about to voluntarily give up the benefits of that context window by outsourcing the thinking that keeps it sharp.

We’ve Been Here Before. And We Mostly Got It Wrong.

If you want a preview of where this ends up, go back to 2008. Technology writer Nicholas Carr wrote a now-famous Atlantic Monthly cover piece called “Is Google Making Us Stupid?” — arguing that the ease of online search was rewiring how we concentrate and think deeply. It blew up. People pushed back hard. The debate raged for years.

Here’s my honest take on that debate, looking back from 2026: both sides were partially right, but they were asking the wrong question.

Google didn’t make us stupider. But it absolutely changed the nature of the critical work required. When you Googled something back in 2005, you’d get ten blue links. You had to evaluate which site was authoritative. You had to triangulate. You learned to recognize the signs of a credible source versus the signs of someone’s angry SEO-optimized blog. The old joke was “skip to page 3 — the first page is all ads.” That joke existed because people were doing the work of curation. It was like going to the library, pulling twelve books off the shelf, and deciding which one to trust. That’s not passive. That’s active, effortful, deeply human reasoning. Even if now the list of links is manipulated by an algo.

The Pew Research Center documented this debate extensively in 2010 — and one of the most interesting takeaways was that the people defending Google argued it would force people to develop better critical thinking because there was simply more information to sift. And for a while, in the right hands, that was even true.

The shift didn’t happen with Google. The shift is happening now.

Google AI Overviews: When the Library Stops Asking You to Read

Pull out your phone and search for something — anything remotely answerable — and there it is: a tidy Gemini-generated paragraph at the top of the results. Google’s AI has already done your thinking. It has synthesized the sources, drawn a conclusion, and presented it to you wrapped in a bow. No links to evaluate. No competing claims to weigh. Just: here is the answer.

A 2025 Pew study found that roughly 18% of all Google searches in March 2025 triggered an AI Overview, and that question-based searches (“who,” “what,” “why”) triggered them 60% of the time. That’s not a niche feature. That’s the default experience for the majority of how people actually phrase searches.

Here’s the thing: it is extremely convenient. I’m not going to pretend it isn’t. I use it. You use it. We all use it. The problem isn’t the convenience. The problem is that convenience is actively removing the step where you had to do critical work.

New research published in 2025 tracking the rise of AI search found that AI-answered search queries surface significantly fewer diverse sources, lower response variety, and in some topic areas, a measurable lean toward particular information ecosystems compared to traditional search. Less variety. Less exposure to competing claims. Less opportunity for you to notice when something smells off – and more opportunity for corporate and government interference in the answer.

And critically — the research showed that people believe AI answers more readily than they believe a random link. The authoritative, confident, grammatically perfect paragraph at the top of the page registers to the human brain as trustworthy in a way that a link to a mid-tier website never did. This is a real problem. The very thing that made Google-era search retain some critical thinking — the requirement to evaluate sources — is being sanded away.

The Research Is In. It Doesn’t Lie.

This isn’t just instinct. The studies are piling up and they’re not subtle.

A January 2025 study published by researcher Michael Gerlich surveyed 666 participants and found a strong negative correlation between frequent AI tool usage and critical thinking ability — mediated specifically by cognitive offloading. When you hand a cognitive task to an AI, you don’t just save time. You deprive yourself of the practice that builds the underlying skill. The muscle doesn’t fire; it doesn’t grow. Younger users (17–25) showed the highest rates of AI dependence and the most pronounced effects.

Microsoft Research and Carnegie Mellon went further in a 2025 study of 319 knowledge workers across 936 real-world AI use cases. The finding that stuck with me: higher confidence in AI was directly correlated with less critical thinking. The more you trust the tool, the less you question it. And the less you question it, the less you actually think.

Seventy-two percent of participants in knowledge tasks reported “much less” or “less” cognitive effort when using AI. Seventy-nine percent in comprehension tasks. Seventy-six percent in synthesis tasks.

Think about that for a second. Synthesis — taking disparate pieces of information and building something coherent — dropped by three quarters. That’s not productivity. That’s atrophy.

When Confabulation Has Real Consequences: The Hall of Shame

Let me give you some examples of what happens when nobody stops to question the AI. Because this isn’t theoretical. People are getting hurt, sanctioned, and embarrassed — at every level of society.

The most famous one you’ve probably heard of. In 2023, New York attorneys Steven Schwartz and Peter LoDuca submitted a legal brief in Mata v. Avianca that cited six federal court decisions. The opposing counsel couldn’t find any of them. Neither could the judge. They didn’t exist — ChatGPT had invented them wholesale: fake case names, fake docket numbers, fake quoted passages, fake procedural history. All of it. When questions arose, Schwartz did the most human possible thing: he asked ChatGPT whether the citations were real. ChatGPT confirmed they were. They weren’t. Both attorneys were sanctioned. The judge called it abandonment of their professional responsibilities. The detail that haunts me is Schwartz’s testimony: “I just was not thinking that the case could be fabricated.” That is cognitive offloading with a $5,000 fine attached.

But it doesn’t stop in courtrooms. Here are three more:

Air Canada’s chatbot told a grieving customer the wrong thing — specifically, it misrepresented the airline’s bereavement fare policy to a customer who had just lost a family member and was trying to book a last-minute flight. The customer relied on what the chatbot said, booked accordingly, and was then denied the discount. Air Canada argued in court that its chatbot was a separate legal entity and it wasn’t responsible for what it said. A Canadian tribunal disagreed. Air Canada lost. The defense of “the AI said it, not us” did not work — and it won’t in the future either.

South Africa withdrew its entire Draft National AI Policy in 2025 — just 17 days after publishing it — because the document cited fake research generated by AI. This was supposed to be a historic moment: the first African nation establishing a formal AI ethics board. Instead it became the first government in history to pull an official policy document because of AI confabulation. The communications minister was direct: the most plausible explanation was AI-generated citations included without verification. “Consequence management” was promised for those responsible.

The Chicago Sun-Times published a summer reading list in May 2025 recommending books that don’t exist — fake titles, fake descriptions, attributed to real authors who never wrote them. Out of 15 books on the list, only 5 were real. The list came from a content partner who admitted to using AI to generate it. People bought the print edition. There was no correcting that.

These aren’t edge cases. A legal scholar’s database tracking AI confabulation in judicial proceedings has catalogued over 200 cases globally — and that’s just the courtroom. The pattern across all of them is identical: someone trusted the confident, fluent, authoritative-sounding output without stopping to verify. The AI didn’t lie to them. It confabulated — exactly like a brain working from flawed ground truth. And nobody applied the critical thinking that would have caught it.

But — Some Muscles You Were Never Going to Flex Anyway

Here’s where I want to complicate my own argument a little, because I’d be a hypocrite not to.

I am never going to learn to code. I’ve made peace with that. I tried. I didn’t enjoy it. I didn’t continue. And because I didn’t continue, I couldn’t build the things I wanted to build — which was genuinely frustrating for years. AI changed that. I am now building things that are real and functional and mine. Here’s a teaser: this whole AI journey — all the reflection, all the late nights, all the agents I’ve babysitted — has quietly turned me into a builder who wants to give something back. I’ve been spending my spare time building something for the rally community. Regional events, TSD, the kind of grassroots racing that most people outside the sport never hear about — not WRC, not national-level anything, just the people who show up on a cold Saturday morning because they love it. It’s coming together in a way that genuinely surprises me, and I can’t wait to share it. Whether it snowballs beyond that little community or stays exactly where it belongs — either outcome is fine by me. Watch this space. The point is: the muscle-I’m-not-flexing argument has a real counterargument. Some muscles you were never going to flex. The question isn’t whether AI is letting you skip work you would have done — it’s whether it’s letting you skip work you needed to do.

And here’s the nuance I don’t see in any of the research papers: even in my AI-assisted building, I’ve been forced to pick up real architecture and structural knowledge. Not syntax. Not line-by-line debugging. But the big stuff — how systems talk to each other, where failure points live, what good design looks like from ten thousand feet. You can’t direct an AI agent well without developing some of that understanding. Which means the cognitive offloading isn’t total, even when you want it to be.

But. And this is a big but.

Prompt engineering is a critical thinking skill. Full stop. I’ve said this before and I’ll keep saying it: it doesn’t matter what part of AI you’re using — better prompts produce better output, period. And in development specifically, this is where the critical thinking stakes get very concrete very fast.

Coding agents, once they get confused on a complex problem, will loop. They’ll circle the same approach, restate the same error in different words, try the same fix from a slightly different angle, and burn your credits doing it. I watch my agents do this regularly on anything sufficiently complex. And the only thing that breaks the loop is a human stepping back, recognizing the pattern, and redirecting — sometimes firmly. “Stop. What you’re doing isn’t working. Let’s look at this from a completely different angle. Forget your last three approaches.” That’s not a technical skill. That’s critical thinking applied to a system that can’t apply it to itself. I’m considering if I can prompt engineer or persona engineer my manager agents to do this for me – I think I can.

Without that ability — without the instinct to recognize when a process has become circular and needs a reset — you will spend days in a rat hole that an AI dug and then helpfully continued digging, deeper and deeper, while confidently assuring you it was making progress. That costs time. It costs credits. And it produces nothing. The human in the loop isn’t optional. The thinking human in the loop definitely isn’t.

We know AI systems confabulate. We know they produce confidently wrong answers with the same smooth delivery as correct ones. We know that the instinct to critically evaluate that output is being trained out of us by the very experience of using these systems. And yet we are deploying them everywhere, for everything, with very little friction between “AI said it” and “I believe it.”

The liability exposure for organizations letting AI-generated answers flow into real decisions without a critical review layer is real. That’s a separate post. But it’s coming.

Bibliographies Were Never Just About Citing Sources

Here’s something I’ve been thinking about that I don’t see anyone talking about enough.

When I was in high school, every essay came with a requirement: a bibliography. A list of every source you referenced. You had to name the book, the author, the edition, the page number. You had to go find those sources, read them, decide if they supported your argument, and then show your work. It felt like busywork at the time. It was not busywork.

That process — finding a source, evaluating whether it was credible, deciding if it actually said what you thought it said, and then correctly attributing it — was a critical thinking exercise disguised as an administrative task. It forced you to triangulate. It forced you to confront the question of authority: who said this, why should I believe them, and does this actually back my claim? It’s almost exactly the same instinct I try to apply when I include authoritative links in these posts. Same idea, different decade.

Now think about what a student does today. They open a chat window, describe the essay topic, and receive five hundred words of confident, grammatically flawless, internally consistent text — with plausible-sounding “sources” that may or may not actually exist in the form described, or at all. The bibliography is still there. The critical thinking that produced it isn’t.

And here’s the part that should bother everyone: plagiarism detection is now effectively broken. For decades, academia fought academic dishonesty with tools like Turnitin — databases that compared submitted work against indexed text to find copied passages. That arms race had a clear enough battleground: human-written text is distinctive, and copying leaves fingerprints. LLM-generated text leaves no fingerprints, because it isn’t copied from anywhere. It’s synthesized. Every output is technically original at the token level even when it’s conceptually hollow.

Worse — even the best AI-detection tools are increasingly unreliable. A student who wants to defeat a detector doesn’t even need to be sophisticated: rephrase a few sentences, shift the tone slightly, change some word choices, regenerate a paragraph. Most detection systems fold. The cat-and-mouse game that took decades to play out with traditional plagiarism has been compressed into a matter of months, and the mouse is currently winning by a country mile.

This isn’t just an academic integrity problem. It’s a signal of something deeper. The reason we required bibliographies and cited sources wasn’t really about catching cheaters. It was about teaching people to build a position on evidence — to understand that an argument is only as strong as the ground truth it stands on. Strip that requirement of its teeth, and you strip the lesson along with it. What you’re left with is a generation that can produce the form of a well-reasoned argument without any of the cognitive infrastructure behind it.

That’s not a skill deficit. That’s an infrastructure failure. And it’s going to compound.

The Education System Has to Step Up. Like, Yesterday.

I’ve been circling this for a while in this series. I said in the Drug Dealers post that broad access to AI matters for humanity. I said in Falling Behind that the pace is creating real anxiety. But I haven’t said this clearly enough yet: the education system has a bigger job right now than it has had in decades, and I’m not sure it knows it yet.

College Board surveys from 2024–2025 found that teachers overwhelmingly worry AI will impede students’ critical thinking development — while a meaningful share of students aren’t even sure what the problem is. That gap is the problem. Students who have grown up with the answer-at-the-top-of-the-search-results experience don’t have a felt sense of what they’re missing. You can’t miss a muscle you never developed.

Even the White House addressed this in April 2025, calling for AI literacy as a foundational skill from kindergarten through post-secondary. That’s good. But I’d argue AI literacy and critical thinking aren’t the same thing — and we need both. Knowing how to use AI well requires knowing how to question AI well, and that’s a critical thinking problem at its core.

Here’s my genuine take: we should be teaching a standalone critical thinking course starting in high school. Not embedded in English. Not a module inside social studies. A real, credited course — call it nothing but Critical Thinking — that teaches source evaluation, logical fallacies, argument structure, identifying confabulation (human and artificial), and the mechanics of how to form a well-reasoned position when everything around you is trying to form it for you. Maybe that sounds old-fashioned. I’d argue it sounds exactly like what 2026 requires.

The Throttle Metaphor (You Knew It Was Coming)

There’s a point in a rally stage — I’ve been there — where you’re approaching a hairpin and the temptation is to keep the throttle buried because it feels like commitment, like speed, like forward motion. The car feels alive. The numbers feel good. But the driver who survives the stage is the one who knows when to lift — who has the situational awareness to recognize that momentum without judgment is just a faster crash.

AI convenience is the throttle. And right now, most of us — me included, I will be honest — are keeping it buried because it feels amazing. Output is flying. Time is compressing. The machine is singing. I wrote about this in the Acceleration Trap post.

But critical thinking is the lift point. It’s not weakness. It’s not inefficiency. It’s the thing that keeps you on the road. And we are, collectively, conditioning ourselves to be worse at it, faster than we’ve ever conditioned ourselves out of any skill before.

The answer isn’t to stop using AI. That ship has sailed. The answer is to be deliberate — almost aggressive — about preserving and rebuilding the habits of skepticism, source-checking, and independent reasoning that convenience wants to replace. Use AI to do the work. Use your brain to decide if the work is right.

The Bottom Line

We laughed at the “Is Google Making Us Stupid?” debate in 2008 and moved on. We mostly got it wrong — not because Google didn’t change cognition, but because the critical layer was still there in the search results. You still had to pick.

That layer is eroding. Fast. A Frontiers in Education study from late 2025 found that students who over-rely on AI for answers show “substantial declines in analytical reasoning capabilities” — and that younger participants are the most vulnerable. This isn’t speculation. It’s in the data.

We are building faster on a foundation that may be getting softer. And the confabulating, confident, extremely smooth AI systems we’ve all become dependent on are both the cause and the reason we’re not noticing.

I don’t have the five-point fix. I never do in this series — you all know that by now. But I do think it starts with admitting the problem out loud. Demanding that our schools, our workplaces, and maybe ourselves take critical thinking seriously again — not as a soft skill, not as a nice-to-have, but as the fundamental human capability that everything else depends on.

Because here’s the thing: AI can confabulate. Humans can too. The only real defense against both is a mind that was trained to ask why before it asks what.

Your turn: Are you still questioning AI answers, or have you found yourself just… trusting them? Did you used to page through search results and pick your sources — and does that feel quaint now? And teachers or parents reading this — do you think a standalone critical thinking course in high school is worth fighting for? Drop it in the comments. I read every one, even if people are not commenting – I’m questioning, did you make it this far? Was it “TLDR” it’s ok, this is how I dump my raw thoughts on random junk – maybe i’ll get back to talking about actually “tech” stuff one day…

(Catch up on the series if you’re new: Falling Behind → AI Is Underpriced → Copilot Price Jump → Drug Dealers → The Acceleration Trap → Time Back → and now this one.)

AI Was Supposed to Give Me My Time Back. Spoiler: It Didn’t. And Honestly, I’m Part of the Problem.

cantechit Uncategorized , , , ,

Let’s do the math out loud.

I’m shipping somewhere between 10x and 20x more output than I was two years ago. Internal tools that would’ve taken a contracted dev team six weeks now take me an afternoon. Research that used to mean three days of tab-hopping is a 20-minute prompt chain. Whole apps — real, working ones — get vibed into existence on a Saturday while my kid eats cereal.

By every productivity logic ever invented, I should be working two days a week.

I’m working closer to seven if you add up work and personal projects.

What the hell happened? AI was supposed to be the thing that finally gave us the time back. Instead it gave us more work to do, more things to chase, more rabbit holes to fall into, more shiny new tools to evaluate, more agents to babysit. The output went up. The hours didn’t go down. If anything, they went up too.

I keep telling myself I’ll write less this week. I keep not doing that. Rathole, I have been asked why I stopped writing – I used to do it more, but as I always say – I’ll write when I feel like I have something to say – lately it’s becoming clear in this series I do, if nothing else it’s cathartic for me to write it down even if I am not sure if someone is listening

Henry Ford Already Proved This Was Possible — In 1926

Here’s the part that drives me a little crazy.

On May 1, 1926, Henry Ford cut his factory workweek from six days to five — 40 hours instead of the usual 50-60 — and kept everyone’s pay the same. He didn’t do it because he was a saint (he absolutely wasn’t). He did it because he’d been quietly studying his own workforce and found something the rest of industry didn’t want to hear: tired workers made more mistakes, had more accidents, and produced less per hour than well-rested ones.

Ford famously said it was “high time to rid ourselves of the notion that leisure for workmen is either ‘lost time’ or a class privilege.” Productivity went UP after the change. Loyalty went up. Turnover went down. Other manufacturers grudgingly followed. Twelve years later the Fair Labor Standards Act made the 40-hour week federal law in the US.

That’s the model. Less hours, same (or more) output, better humans. Proven a century ago with a stopwatch and a Model T assembly line.

So now, in 2026, when I have a tool that legitimately makes me 10-20x faster at most of what I do — by the Henry Ford logic, my workweek should be collapsing. Not just to four days. To two.

And it’s not.

Why It’s Not Happening (And Won’t)

I went looking for the answer in a few places. It’s not one thing. It’s a stack.

Microsoft’s own data is the kicker. Their 2025 Work Trend Index surveyed 31,000 workers across 31 countries and crunched trillions of M365 productivity signals. The headline result: knowledge workers are now interrupted on average every two minutes — 275 times a day. This is entirely my life, I could write entire articles about how I don’t do my best work when I am task switching. One in three say the pace of the last five years has made it impossible to keep up. 53% of leaders demand productivity increase; 80% of the workforce reports they don’t have the time or energy to do effective work. Microsoft literally calls it the “Infinite Workday”. Their own warning is brutal and quotable: we risk using AI to accelerate a broken system.

Which is exactly what’s happening. We didn’t redesign the workday around AI. We just bolted AI onto the existing infinite one and made the meter spin faster.

Then there’s Jevons. I mentioned this in my Cisco AI Summit post — Sam Altman flagged it on stage. The Jevons Paradox is the 19th-century observation that when something becomes more efficient to use, we don’t use less of it; we use way more. Coal got cheaper, coal consumption exploded. Compute got cheaper, compute consumption exploded. Now AI is making knowledge work cheaper, and Box CEO Aaron Levie went viral last December arguing the same thing applies: cheaper intelligence won’t shrink work, it’ll expand it. The bar for “what counts as a deliverable” just keeps moving up.

French consultant Bertrand Duperrin nailed it in a February 2026 piece on AI work intensification: if nothing stops, then everything gets added. AI becomes a machine for densifying, for multiplying iterations, for expanding the scope of what’s required. Less a transformation of work than an intensification of it.

That’s me. That’s exactly me. I’m not doing less; I’m doing the same five-day week with five times the surface area.

And the historical pattern says don’t expect this to change. Back in 1930, John Maynard Keynes wrote a famous essay called “Economic Possibilities for our Grandchildren” predicting a 15-hour workweek by 2030. He wasn’t crazy — he just assumed humans would bank productivity gains into leisure. A 2022 LSE revisit of the prediction found the productivity gains absolutely happened (real GDP per capita more than quadrupled), but we didn’t take them as time off. We took them as more stuff, bigger houses, longer retirements. Lifetime leisure DID rise about 58% in the UK — but almost all of it from people living longer in retirement, not from working less while employed.

We’re four years from Keynes’ 2030 deadline. The average full-time worker is nowhere near 15 hours. The Henry Ford 40-hour week, set up for factory floors a hundred years ago, is somehow still the ceiling and the floor for knowledge work designed for digital agents.

So the world had its chance — multiple chances — and chose more consumption over more time, every single time.

The Confession: I Could. I Won’t.

Here’s the part where I have to be honest, because this whole post would be a cop-out otherwise.

I’m not just a victim of the system. I’m one of the people running the meter.

Could I do my actual “value” work in two days and then sit on my hands for three? Honestly, probably yes. METR’s May 2026 survey of technical workers puts the median self-reported speed change from AI at 3x and median value-of-work change at 1.4–2x. I’m self-aware enough to know I’m above the median — building with multiple agents in parallel, vibe-coding tools in hours that used to take weeks. The math is the math. I could compress.

But I won’t. And not because some boss is making me — I’m in a fairly senior role and I have a lot of control over how I spend my hours. The reasons are uglier and more personal than that:

  1. I’m addicted to the capability. I wrote about this in AI Companies Are Straight-Up Drug Dealers. When the billing glitch killed my agentic access for two days I felt actual withdrawal. Sitting on my hands for three days a week while I have a Claude Opus session sitting RIGHT THERE that could be building something? Not happening.
  2. The competition resets the bar instantly. I wrote about this too in Falling Behind. The second I deliver an internal tool in two days that used to take six weeks, that becomes the new normal. Next week’s ask is bigger. There’s no going back to the slower pace because the world I work in already adjusted to the faster one.
  3. I want to build cool stuff. Two days a week of work means three more days of “what should we build today?” That’s not rest. That’s just a different kind of building. I’m also building my own stuff on my hours and days off.. Disconnecting is harder than ever.
  4. My brain is buzzing along with the GPU. I wrote about this in The AI Acceleration Trap — the VRM hum, the insomnia, the skipped meals. I KNOW it’s costing me. I keep going anyway. That’s not a productivity story. That’s an addiction story dressed up in productivity clothes.

And here’s the kicker — there’s a 12-person software startup called Convictional that actually did the Henry Ford move in mid-2025. Moved to a 32-hour, four-day workweek. No pay cut. They credited AI absorbing enough manual work to make it possible. One company. Twelve people. That’s the entire counter-evidence I could find. Nobody else is volunteering this. Bosses won’t. Boards won’t. Investors won’t. Jevons won’t. And honestly? Neither will I. If I do it, someone else will be glad to keep the 40 hour.

So What Do We Do? Honestly… Probably Nothing.

This is the part where I’m supposed to give you the five-point action plan to reclaim your time, and I’m not going to insult you with one. I don’t have one. The 5-day, 40-hour week survived the Industrial Revolution, electrification, the PC, the spreadsheet, the smartphone, the internet, the gig economy, and a global pandemic. It’s not going to fall to AI either. The only thing that changes is how much we cram into those five days.

It’s a little like the line in rally — there’s a place on every stage where you KNOW the smart move is to lift off the throttle, scrub some speed, and bank a clean exit. And yet the same drivers who know that, including me on Fire Access Road 507 with absolutely no prize money on the line, keep their foot in it. Because the throttle feels good and lifting feels like losing. The math says lift. The hands keep pressing.

So I’m not lifting. I’ll aim for the traction from the ruts like Crazy Leo taught m, I’m probably going to keep pushing 6-7 days a week. I’ll keep my guardrails from the Acceleration Trap post — protect sleep, eat actual meals, take walks, keep the agents on a leash so they don’t burn through credits or my nervous system overnight. But the workweek itself? It’s not getting shorter for me, and unless your boss is Henry Ford reincarnated as a 2026 CEO with a stopwatch and a conscience, it’s probably not getting shorter for you either.

May as well enjoy the ride. Just don’t ignore the buzzing.

Your turn: Are you doing 5-10x the work in the same hours and getting away with it, or have you actually carved out time back? Did your employer give it to you, or did you take it? Or are you like me — could compress your week to two days and refusing to, because the dopamine and the FOMO are louder than the math? Drop it in the comments. I’m reading every one.

(Catch up on the AI builder series if you’re new: Falling Behind → AI is Underpriced (But Not For Long) → GitHub Copilot Price Jump → Drug Dealers → The AI Acceleration Trap → and now this one.)

I feel am falling behind – but so do many.

cantechit Uncategorized , , , ,

Right now, sitting here staring at my screen on a random Thursday morning in April 2026, I feel like I’m falling behind.

This is supposed to be the next post in my AI series – the one where I keep talking about vibe coding, turning coders into builders, and (more importantly) turning non-coders like me into builders and innovators. But today? Today this isn’t going to be another “here’s how I hacked something cool with AI” post. This is going to be an honest, chatty, let’s-be-real moment. I am tired of pretending I’ve got it all figured out. It’s changing faster than I can learn it – and yeah I am feeling some FOMO.

I’ve never felt comfortable in front of an IDE. Never. I open VS Code and my mind goes blank. I’m bad at syntax, the terminal throws errors that feel frustrating, and half the time I’m just copy-pasting whatever the AI spits out and praying it doesn’t explode in project, thankfully never really production. I can look at some of the code I’ve “built” lately and straight-up tell you: some of it is complete garbage. It’s messy. It’s inefficient. It breaks every best-practice rule in the book. I hate saying “Look at this thing I wrote” because I didn’t. I also hate the term “I Vibe’d it” – I’m using “Build” for now.

But you know what? It works.

And for me… sometimes that’s enough.

That’s the dirty little secret I don’t see a lot of people admitting out loud in this whole vibe-coding wave. We’re out here describing what we want in plain English, hitting enter, and watching magic happen. No hand-written algorithms. Just vibe. And yeah, it gets stuff done faster than I ever could on my own. But it also leaves me feeling like a hack. Like I’m riding a rocket ship I didn’t build and don’t fully understand. Thankfully no human lives on the line here.

And then I look at my friends.

Take the great John Capobianco. That guy is constantly vibing entire projects into existence. He’s out there building VibeOps communities, spinning up AI agents that feel alive, turning weekends into prototypes that actually ship. I watch what he and others are doing and I’m genuinely inspired… but I’m also hit with that punch-to-the-gut feeling: “Damn, the world just moved on again.” I finally get comfortable with GitHub Copilot and suddenly everyone’s talking about the next thing. I learn one new trick and three more drop that make it feel obsolete. It’s exhausting. Insert GooberClaw or whatever new “Claw” is out this week, I have yet to even try John’s NetClaw because frankly some of those things I just don’t trust – but somehow in a dumb way I trust my own vibe coded stuff – that’s pretty dumb.

I’m not alone in this. NetworkChuck dropped a video the other day called “I kind of hate AI… and it almost made me quit YouTube.” I watched the whole thing and just nodded the entire time. He straight-up says it: it’s a love-hate situation. The pace is relentless. Even on sabbatical he couldn’t escape it – AI was in his feed, in his conversations, in his head 24/7. He felt paralyzed. He hated that he hated it. I felt that in my bones. You nailed it Chuck.

Here’s the thing nobody talks about enough: this speed is creating real stress and anxiety.

Reaching out ot my AI Friends to help me research this one…. University studies back it up. A 2020 study by Rosenstein, Raghu, and Porter at UC San Diego (published at SIGCSE ’20) found that 57% of computer science students experience frequent impostor feelings – 52% of men and a whopping 71% of women. That was before the AI explosion. Fast-forward to 2025-2026 and a new Eastern Washington University survey of 1,000 workers shows that people using AI daily are the most likely to report regular impostor syndrome (30%). Another Ernst & Young study found 66% of employees are anxious about falling behind if they don’t use AI, and 65% are stressed about not knowing how to use it ethically.

There’s even a term for it now – technostress – and research in PMC shows AI-generated technostress indirectly tanks quality of life through spikes in negative emotions. We’re all feeling it: the pressure to keep up, the fear that if you blink you’re obsolete, the quiet voice whispering “you’re not a real builder.”

I feel that voice every single day.

But here’s the flip side – the reason I’m still writing this series and still showing up.

Vibe coding isn’t just about perfect code. It’s about democratizing building. It’s about taking non-devs like me and saying, “You don’t need to be fluent in three languages and have 10 years of LeetCode problems solved. You can describe the problem, iterate fast, and ship something that moves the needle.” It turns coders into faster builders and non-coders into innovators who never would have started.

I’m also taking steps to get others on the train and mentor others, my mentors have recently said to me “I won’t be here for ever, it’s time you start doing more” <– I am embracing this. Just yesterday a colleague talked to me about how he wished he could build something – I asked him if he had tried, then showed him 60 seconds of what’s possible – he was excited to start, and went on his way. Maybe I am more ahead than I think – there’s that impostor thing again.

Some of my code is garbage? Cool. It solved the problem in my lab in an afternoon instead of a week. John is out there vibing entire platforms into existence? Amazing – I’ll keep learning from him and cheering him on. NetworkChuck is honest about the hate part? Respect – it makes the love feel more real. These experts inspire me.

So yeah… I feel behind. I feel insecure. I feel the anxiety of “doing well” in a world that doesn’t slow down. But I’m also still here, still experimenting, still believing that “it works” is a valid starting point when you’re a systems guy who never planned on being a builder.

If you’re a non-dev reading this and you feel the same way – welcome to the club. If you’re a dev watching the vibe-coders and feeling a bit of whiplash – you’re not alone either. If you think the stuff we are building is bloated trash — you are 90% right – sometimes.

Drop a comment. Tell me where you’re at. Are you riding the wave or white-knuckling it? Let’s keep the conversation real.

At the end of the day, vibe coding was never about being the best coder in the room.

It was about giving more of us a seat at the builder’s table.

And I’m still showing up to that table – garbage code and all, still feeling out of place, still feeling behind.

(And if you’re new here, catch up on the AI series: Vibe Coding with GitHub Copilot and OpenClaw: The Passion-Driven AI Agent.)

Cisco Live 2025: Community, Balance, and Big Dreams for AI

cantechit Uncategorized , , , ,

That was huge.

Still buzzing from Cisco Live 2025 in San Diego. This wasn’t just a tech conference—it was a reunion of brilliant minds and big hearts. The Cisco Community Champions dropped wisdom that flipped my perspective, like a breakout session chat that rewired how I think about collaboration. Tech Field Day delegates brought the heat over late-night tacos, debating tech’s future with ideas that stuck with me. And my Cisco colleagues and friends? They’re family—coffee in the DevNet Zone, laughs at the Customer Appreciation Event (The Killers absolutely slayed!), and moments that recharge my soul. But as I look ahead, I’m thinking about balance, mentorship, and how we’ll make a real difference with AI. Here’s the vibe.

The Community That Fuels Us

Cisco Live is all about connection. Those conversations with Champions, delegates, and friends aren’t just chats—they’re sparks that ignite new ideas. A mentor’s advice over drinks is already shaping my next move, and the energy at the CAE was pure magic. This community pushes us to dream bigger and work smarter, together. Those that challenge me, thank you for doing that. The support from those who love our work challenges us to do more. Without community – we really have nothing.

What’s Next for Me: Building, Mentoring, and Balance

This year, I’m all about building. I’m diving into relationship building, leveling up my skills in innovative problem-solving, and finding new ways to share with you all. I’m hyped to get back to blogging, maybe even start vlogging, but I’m keeping it real—it’s a lot. Tech can be a grind, and we don’t always talk about the psychological toll it takes. The pressure to stay ahead, the endless hustle—it weighs on us. I’m prioritizing balance, making time for myself, and I invite you to do the same. Check in on your friends and colleagues, be that supportive ear. We’re stronger when we lift each other up.

I’m also building a structured mentorship plan to guide others, inspired by my own mentors. Whether it’s sharing tech insights or navigating career challenges, I want to pay it forward and help others shine. Who knew my greatest challenge by my own mentors, would be to pay it forward. I have started to realize and understand that climbing this career mountain hits a plateau, and unless you can lead a team to the top – you are now stuck.

Making a Difference with AI and Country Digital Acceleration

This year, I’m wrestling with a big question: How will I make a meaningful difference with AI? It’s consuming my thoughts. AI has so much marketing hype – I want to get past that. AI has the power to transform lives—think smarter cities, safer communities, inclusive access to tech – or just making super complex things – easier. At Cisco Innovation Labs, we’re celebrating 10 years and the anniversary of Country Digital Acceleration (CDA). I’m so grateful for CDA’s support, backing projects like Digital Canopy that bring connectivity and hope to underserved areas. Their belief in our ideas fuels us, and I’m stoked to deepen our work together, dreaming up solutions that change the world. This is a great partnership, and it really gives us the ability to “Design with Empthy, Innovate with Purpose”

The Next Big Thing for Our Labs

With a decade in the rearview, it’s time to go big. What’s the next big thing for Cisco Innovation Labs? I’m obsessed with figuring this out. Maybe it’s AI-driven public safety tools, or … well… so many things I can’t talk about yet… , or sustainable tech that powers a greener future. Whatever it is, it’ll be bold, human-centered, and built with this incredible community. I’m ready to dream, experiment, and make waves. I know one thing, technology comes second, people, community and EMPTHY come first.

Keep the Vibe Going

Cisco Live 2025 was a love letter to community, a reminder to stay connected and take care of ourselves. As I chase big dreams with AI and our Labs, I’m carrying this energy forward. So, take a moment for you, check in on your people, and let’s dream big together. What’s your Cisco Live highlight? Hit me up on Twitter or drop a comment—let’s keep it rolling!