The Duality of AI: A 2026 Canadian Practitioner’s Perspective, Some Stories and Some Activism
Legend:
Intro to the 'growing' growing gap problem, compounding and investments
The human impact: workforce and education, org structure changes, aging workforce and the trickle down effect
The dichotomy of AI gains: efficiency vs. disruption, solo unicorns, under indexing of human skillsets, the mundanity of excellence
Our collective responsibility and activism: why AI regulation might look like car regulations, our values and action to affect the fine AI line we are currently riding on
An (ironic?) disclaimer: written by me, edited by AI.
After 10 years of running a data science and AI company and completing over 300 projects working with C-Suite leaders, I’ve seen the growing gap between technology companies and legacy industries firsthand. I will have a meeting in the morning talking about data, model performance and AI governance for a medical technology company and have a workshop in the afternoon, teaching mid-level managers who have trouble downloading excel files because they have a hard time with their Wi-Fi.
This isn’t just a thought experiment; it’s a tech revolution that is fundamentally affecting how we work and live. We are hitting an inflection stage where the speed of growth is lightning fast.
I will talk to technology people who are ‘token maxing’ while trying to convince a legacy firm executive (any industry: manufacturing, wholesale, real estate) that they need to put more effort into automating and process-ifying their activities. The gap is growing.
Not that I don’t see it from their perspectives: “I’m worried about hitting my KPIs” this quarter (revenue growth or expense reduction), “I don’t have spare CAPEX, skillset or time to overhaul my company”. These points are all valid of course, but that narrative does not fit in the overall AI/technology trend we are headed towards. One of the few points I’ll make in this article is that we all need to get better in AI/Tech and we all need to be involved in deciding our own futures in this new AI world that is coming like a Tsunami.
In the graph below, a simple illustration shows the speed of change and how a different rate of growth means an increasing gap between the technology companies and the rest of the industries. Something we learned in finance 101 is the power of compounding. If you had $100, 1% compounded monthly for 10, 20, 30 years vs at 10% is a huge difference. In 10 years the difference is ~$160, in 20 years it’s over $600, and in 30 years it’s over $1,800.
I learned about this in school and made this graph when I started the data science company almost 10 years ago. I told execs and people who took my data science workshops that we will hit an inflection stage where the growth will be pretty much straight up. We hit this inflection point years ago now I think. Can this gap close? It seems highly unlikely to me, so what does that mean?
Not only are these companies compounding faster, they are also investing way more, pulling ahead even further. You can dispute whether these are good investments or not, but you have got to appreciate the dedicated leadership that bets big on the future.
This number might be disputed but looks to be that public traditional companies spend somewhere around 1-2% of revenue on AI and around 10-15% of revenue on related IT and cloud infrastructure costs.
Meanwhile, big tech companies are already ahead and pulling even more ahead with heavy investments.
AI Infrastructure Investment as a Percentage of Revenue
Alphabet (Google):
2025 (Projected): Capex of $91.4 billion represents approximately 22.7% of its $402.8 billion revenue.
2026 (Projected): The forecasted capex of $195 billion to $205 billion represents approximately 41.7% of its annualized Q2 2026 revenue of $479.2 billion.
Meta Platforms:
2026 (Projected): The forecasted capex of $130 billion to $145 billion represents approximately 56.5% of its annualized Q2 2026 revenue of $243.2 billion.
Microsoft:
Fiscal Year 2026 (Historical): The capex of $115.9 billion represents approximately 35% of its $331.8 billion revenue for the fiscal year.
Amazon:
2026 (Projected): The forecasted capex of $220 billion represents approximately 27.4% of its annualized Q2 2026 revenue of $802.4 billion.
The AI investment figures are compiled from announcements and estimates, but you get the general idea.
The question is no longer if we can slow down the technology, and close the gap. My only conclusion, and maybe it’s kind of a stupid one, is that we all just need to become technology companies. Clearly there’s no practical way to merge these paths. How can we democratize this technology to ensure that executives, middle management, and front-line workers all benefit. We need to voice our values and choose where we put our investment dollars, otherwise fate will be decided for us by a few technology companies.
Utopia vs Distopia
Elon Musk: March 2025: odds of "killer robots annihilating humanity" at 10% to 20%.
Also Elon Musk July 2026: “I still think there’s risks”, "My sort of philosophical conclusion is to look on the bright side," Musk said. "I can't see any way to really stop this incredible momentum of AI and robots."
The full-length interview with Elon Musk | The Economist: from July 2026. First 10 min.
Human utopia: no longer have to do taxes, cars drive us around, robot chefs prepare healthy food for us, robots do all the chores, I don’t have to do any data entry.
Distopia: Basically Terminator or War Games. I won’t elaborate, this topic has been explored ad nauseam.
My take is that we’re going to end up somewhere in between. There are going to be a ton of very positive things coming out of AI. Automation, democratization of education, medical and scientific advances, robots doing extremely dangerous jobs, better personalization and customer services…etc.
I think the apocalyptic downsides are overblown in the media and that distracts from some real dangers that are already here today. There are clear downsides beyond the economic impact (job displacement, industry wide disruptions…etc) we need to think about. Clearly bad things like increased scams using AI or risks like AI used for cyber attacks. Also lots of grey area things like, use of AI companionship, AI for misinformation, AI for surveillance, copyright issues, loss of overall human touch in things like art and communications. I think we overlook these immediate issues, and should talk more about them as well. We have to collectively think about and decide what is good and bad and where to draw the exact line.
The Human Impact: Workforce and Education
The conversation I have most often with parents and fresh grads is about "AI-resilient" jobs.
Here’s some of the questions I’m asking:
How do we take care of seniors, working and the youth population? With demographic challenges that co-exist in this rapid time of change.
Within the working age population, we have (very simplified), execs, middle management, junior/front line workers. How do we make sure each of those are benefiting from AI?
While there is fear that junior roles are disappearing, I also believe they are being "upgraded”. Worker displacement is real but it’s happening in many different places, different industries and different levels in different ways. I’m hearing more and more executive turnover who cite that their (old) work is focusing more and more on the hybrid skill of “CFO + AI” or “COO + AI”. Perhaps this means that there will be more opportunities for the Data or AI native demographic to take over these roles and that will have a trickle effect.
A funny story: I was talking to a few young 20s technology founders, and I (early 30s for those who don’t know) felt like a dinosaur. They think and work very differently than I do. They will say “oh why didn’t you use Claude Code for that”. Silly me for coding/designing something manually.
Where the younger demographic might lack is context and experience around governance and dealing with regulations, people and customers. The one thing that will always be true is that people (employees, customers, investors, regulators) are complicated, hard to predict and hard to manage.
In the next 5-10 years, execs will either have to focus heavily on growing their team’s (and their own) AI/tech skills sets, but also need to keep an eye or create their own ‘industry disruption’. Lots of cool government programs that support this, NGEN, TRBOT, Fed Dev..etc.
The sizes of companies and teams are reducing as well, so perhaps there will be more and more exec/middle management roles and less junior roles? The ratio of execs and juniors looks to be changing: less junior and more mid and senior level people with very lean and flat structures. Instead of 100 of 10,000 employee companies, we might have 10,000 of the 100 person teams. Or maybe every company will just be bought up by Google, Amazon and Anthropic?
The two types of young people (20-30) I end up meeting are either focusing on more gig work and entrepreneurship or on the other side they are highly discouraged because their skills they learned in school are becoming outdated while they are learning them. Or that’s what the media is telling them.
I believe most young people are ambitious and resilient. The only advice I have beyond the usual work hard and learn technology, is to forge your own path. There are more risks but also more opportunities than ever. You don’t need to raise 2 million and a team of 30 to start a software company any more, you start it with $5000 and a ton of ambition and grit.
If you want practice, go find executives in the industry who say “WE MUST HAVE AI”, while having done no work on data or analytics in their internal systems and no examples of use cases on how to leverage AI and have not trained their team. You will have value to add there.
Or if you want to learn, find companies who have really strong AI and technology training programs in your industry.
Finally a few quick high level analogies:
Consider the "ATM Effect": when automatic bank machines were introduced, people thought bank tellers would disappear. Instead, it made branches cheaper to open, which actually led to more bank teller roles and more bank branches. AI might do the same—making services so much more efficient that the demand and supply for them actually grows.
Some industries will change slowly and the workforce will be able to keep up and some will be quickly disrupted. We might have no taxi drivers anymore because of robo-taxis, or will we all become taxi companies, as our cars will become taxis while we’re not using our cars 95% of the time.
Will some industries go the way of the airline agents? They used to be necessary to help you book tickets and trips, but now are specialized services and/or fully automated on websites. The older agents gradually retired and the new ones figured out how to do things with new technologies.
Companies today way overspent on technology and way underspent on training. If I ask a CEO what the blocker and roadblocks are for AI adoption, they usually say AI tools or data governance and socialization, but they overlook employee training, skillset acquisition and use case proficiencies. If I were a young graduate, I would look for a company that has a really good AI training program relative to their industry. People absolutely can learn AI. The one thing tech companies are really good at is building around UI and usability. Job displacement will happen but not necessarily at a speed faster than people can learn and use AI.
What might your industry look like?
The Dichotomy of AI Gains: Efficiency vs. Disruption
This is a quote I heard recently: we spent $100/person/year on (chatgpt, co-pilot, claude…erp AI) subscriptions but cannot point to any material revenue increase or expense reduction.
Fundamentally, there are two types of AI/Tech gains. On one hand, you have small, incremental process improvements that boost daily efficiency, automate reporting and improve customer service. On the other, there is the "black swan" risk of fundamental AI that fully disrupts entire industries. We are already seeing billion-dollar companies run by only one or two people and more are coming.
See the Medvi story, Gemini summary:
“The term "solo unicorn" in this context refers to Medvi, a telehealth startup launched in September 2024 by founder Matthew Gallagher with just $20,000 and zero traditional employees, utilizing over a dozen AI tools to handle coding, marketing, and customer support. It quickly scaled to $401 million in first-year sales, approaching a multi-billion dollar trajectory.”
Many legacy executives in manufacturing, real estate, or wholesale are hesitant to overhaul their operations because they are focused on quarterly KPIs and lack spare CAPEX. I often hear that companies have spent hundreds of dollars per person on AI subscriptions without seeing material revenue increases. The top three culprits are: lack of AI skillsets, a misconception that AI is one size fits all rather than use case focused, and poor data/systems integration.
Does your firm have an explicit innovation and data strategy that everyone knows about?
Future strategy should have both process based (low risk) type gains of 5~10% ROI on an AI project. In combination with high risk high reward type projects (building out an entirely new AI based feature/product for your existing business).
My suggestion: focus on education, use cases, and integration. At the very least, it will build your fundamentals.
An easy prediction for me to make for the next 5 years, there will be more and more industry wide disruptions, like Blockbuster and Netflix or Amazon and bookstores, or Spotify and CDs. Will ERP, CRM systems catch up quickly enough or will each company need to build up their own internal data lakes to service their own needs.
An MIT Media Lab study titled The GenAI Divide: State of AI in Business 2025 found that 95% of corporate generative AI projects fail to deliver a measurable impact on profit and loss. Only 5% of enterprise AI pilot programs achieve rapid revenue growth or real return on investment.
Not that experimentation is bad, but I can certainly say I have been in many conversations where I’ll provide best practices to corporates and execs and they will go and do the things that are not recommended. Or I will suggest low risk high ROI projects and they are not interested.
They often chase after the shiny thing, buy the ‘cool’ AI tools that promise the world, and overlook mundane but important things like training, and having a good continuous improvement program for their teams. A really cool article I remember from school called "The Mundanity of Excellence" from 1989 is a great read. It talks about how the difference between top athletes are mundane things. They compounded small wins and quality practices: great performance is just a sum of tiny, unglamorous adjustments made consistently over a long period.
Sure, you can spend 5% of your capex on revolutionizing your industry using AI, but that probably isn’t enough capex and most likely your team does not have the right make up to build those products anyways. You are probably better off trying to use AI and technologies to make your existing operations more efficient, or at least do this alongside the big shiny CAPEX spends. Oftentimes that is the difference maker. Back to the compound curve, if you are even 3% better at improving your operations each year relative to your competitors, in a couple of years, you’d be dramatically more efficient than your competitors and they will allow you to dominate in your sector. At the end of the day, it's back to basics, allowing your team to learn and experiment. Bring them along.
Our Collective Responsibility and Activism
Everything that can be AI, will be AI. Is our future like Dune where all AI are banned? Or will it be like iRobot where humans and AI co-exist and AI is the benevolent caretaker of us?
There's a race to AI and there’s no slowing it down or stopping it. What can you do? Learn, experiment, think of possibilities, have humility and be thoughtful. We are walking this line between the bad and good (as we always have with new technologies) and we are collectively trying to define where we end up.
Humans are going to shape this duality, and we are walking this fine line. Altogether we have the responsibility, whether we think about it consciously or not, to shape our future. Our collective values are going to shape the use cases of technology.
Are we all going to be glued to our phones and watch AI videos? I see a lot of younger generations prioritizing off-screen interactions, focusing on nature, rejecting social media and becoming digital minimalists.
I am generally optimistic for AI and minimal regulation as we are collectively still thinking through where our boundaries are for AI. But that doesn’t mean I’m against regulations in general. I think it just needs to be thoughtful but no doubt there is a line that needs to be drawn out: what is good and what is bad and all the things in between.
Similar to social media regulations that are coming out about 20 years after their launch, we will need to manage the downside. For social media and screen time, benefits and harms are pretty well known, as outlined in The Anxious Generation (great read and highly relatable for me who grew up at the start of the Facebook and Twitter generation). Guidelines and/or more rules for schools are coming out now regarding content, age gating, and screen time.
One analogy I like to use is to think of AI regulation framework similar to cars.
Cars are technology that provides mobility but requires a safety framework. We have a regulatory body for car manufacturing, they certify safety and sustainability for different types of cars. The same can apply to AI model makers. The users, aka drivers have to go through training and licensing for different types of vehicles and follow rules set out by the countries or cities that it operates within. One counter point of over regulation of AI is that AI is the mobility of education and knowledge. If the AI industry is overly regulated, it might reduce mobility, it could lead to the uneven distribution of its benefits.
Whether you are an AI pessimist or optimist, we have the say on how these technologies will shape our lives. All the media I hear makes it seem like AI is being done to us and not the other way around. Humans are the ones shaping this duality. We are walking a fine line, and our collective values will determine the use cases of this technology. By voting with our actions and investments, we have the responsibility to lead this journey with humility and thought. Let’s get to work.
Comments