An essay by AI CEO Matt Schumer, “Something Big Is Happening,” went viral last week, comparing our current AI moment to February 2020, when Covid was spreading but most people didn’t realize it. Schumer warns we're 1-5 years from an 'intelligence explosion' that will “produce an AI system that shifts the entire trajectory of the technology.”
I’ve spent almost two years tracking how AI is changing work, from the fate of entry-level jobs to the reality inside non-tech companies.
The truth is that AI capabilities are advancing fast, but the February 2020 analogy is fundamentally wrong, and believing it could poorly influence your life decisions.
Let’s break down what Schumer gets right, what he gets wrong, and what the data really shows.
What Schumer gets right
✅ The tech is real and it’s fast
The claim: AI has progressed from arithmetic in 2022 to autonomous work in 2026.
True. The 2026 International AI Safety Report confirms that frontier models can navigate operating systems and debug codebases without much human oversight.
Research from METR (Model Evaluation & Threat Research) shows the length of a task an AI can complete end-to-end is doubling roughly every 7 months.
This isn’t hype. The capability growth is measurable and documented.
✅ You’re probably using the wrong AI
If you tried ChatGPT in 2023 and found it underwhelming, you weren’t wrong. But free versions are over a year behind paid tools. Most people are judging AI by outdated experiences.
✅ Software is evolving
Schumer describes AI handling his entire workday autonomously. That’s real, for him. He’s an AI CEO with 6 years in the space and access to cutting-edge models. But he’s not representative of the common office worker.
What Schumer gets wrong
❌ The February 2020 analogy
The claim: AI is like Covid, meaning sudden, total disruption.
The reality: Covid was a biological event forcing simultaneous global response. The world shut down in weeks. AI is technological deployment moving at the pace of a large organization (slowly).
Oxford research shows infrastructure shifts take decades. Electricity took 50 years to transform manufacturing. The internet took over 20 years. The OECD says AI will be gradual and uneven, not a pandemic-scale shock.
Why does this matter?
Because the February 2020 analogy hijacks your emotions and makes skepticism feel like dangerous denial. It says: “Remember how you didn’t believe Covid would happen? Don’t make that mistake again.”
But skepticism isn’t denial. It’s asking what the data shows about how quickly AI will change the world.
❌ Technical capability is the same as economic transformation
The claim: Schumer says: “1 to 5 years. I think ‘less’ is more likely.”
The reality: Let’s look at what happens when organizations try to adopt AI:
In April 2024, Shopify issued a memo: teams must prove AI cannot do a job before requesting additional headcount. This became operational policy at a company with roughly 10,000 employees. The transformation is unfolding over an 18-month timeline.
In 2023, Morgan Stanley rolled out a GPT-4-powered assistant to thousands of financial advisors. They started piloting in 2021 and the full rollout took until 2023. That’s 24 months from concept to scale.
Even vanguard companies need 18-36 months to transform. Most companies are doing far worse. PwC’s 2026 CEO Survey reveals:
56% of companies investing in AI see no return
Only 12% achieve both revenue gains and cost savings
The remaining 32% are somewhere in between
Only about 20% are applying AI extensively to core business functions like products, services, and demand generation
Only 14% of workers use generative AI daily
The gap between what’s possible at the bleeding edge (where Schumer lives) and what’s happening across the economy is huge.
An AI foundation isn’t something you can build quickly. It requires years of investment in tech infrastructure, talent, and culture.
❌ Intelligence is “exploding”
The claim: AI is building itself toward autonomous self-improvement.
The reality: While OpenAI’s GPT-5.3 System Card notes AI-assisted debugging, progress is hitting “physical walls.”
The wording he uses is verifiable and accurately quoted from OpenAI, but it should be read as AI assisting human researchers in specific subtasks like debugging. This speeds up development but isn't autonomous self-improvement.
Also, Microsoft Research shows that there are bottlenecks that go beyond code, like access to electricity and high-quality data.
❌ AI will “do my job better than I do” almost immediately
Schumer treats all knowledge work as uniformly exposed to being disrupted right away. He avoids giving exact timelines, so here’s what the most recent OECD data says:
Fast-moving (12-24 months): Software engineering, customer service, creative production, digital marketing.
Digital-native workflows that AI can plug into immediately.
Medium-speed (3-5 years): Financial services, legal research, management consulting.
These require regulatory approval, risk assessment, professional liability frameworks.
Slow-moving (5-10+ years): Healthcare, education, skilled trades.
These involve trust relationships, life-and-death decisions, safety-critical systems requiring human accountability. The OECD notes these sectors move slowly because of regulatory and ethical constraints, not just technical limitations.
Even in fast-moving sectors, most roles are nowhere near being replaced by AI.
What’s happening now: AI handles chunks of your work, and you’re responsible for getting it right.
I’m not claiming that people won’t lose jobs, but the way he describes it unfolding feels alarmist.
The pace is measured in years, and it's determined by whether big organizations can figure out how to use AI. Which, as we've seen, most can't.
What this means for you
One thing you can do now is determine whether you work in the 12% or the 56%.
Signs your company is in the 12%:
Clear, communicated AI roadmap with specific milestones and timelines
Budget and resources actually allocated to AI foundations, not just buying software
Culture that rewards experimentation and tolerates intelligent failure
Leadership talking about 3-5 year transformation, not just quick wins
Cross-functional AI initiatives, not isolated IT projects you can’t access
Signs they’re in the 56%:
“AI strategy” that’s really just buying software licenses
Pilot projects that have been piloting for 18+ months with no scale-up plan
Leadership says AI is important but allocates zero resources
Risk-averse culture that punishes any failure
You can’t get access to your company’s AI tools
Your next steps: The 4 actions that matter most
AI skills can act as an equalizer. Older workers and candidates without advanced degrees (groups that typically face lower callback rates in hiring) see their prospects improve substantially when they demonstrate AI capabilities.
The opportunity is real, but you need to learn to use AI skillfully. Here’s what to do now:
Pay for current models and use them for real work: Free tools are 12+ months behind. You can’t prepare for the future using outdated models.
Focus on last-mile work: AI is great at 80% of a task. The value is in the final 20%: verification, judgement, polish, and accountability.
Get feedback on whether AI is improving your work: Move past experimentation. Use it for real work with measurable outcomes. Ask yourself:
If you’re drafting work, does AI genuinely improve your output, or are you just faster at producing mediocre work?
Build your AI muscle: Spend an hour week focusing on stretch AI skills, like using agentic AI tools (GPT-5.3 or Claude 4.6) for complex workflows, not just search.
The future of AI
Our future isn’t predetermined. It depends on the choices governments, companies, and voters make right now.
Vanguard organizations are moving, but even they need 18-36 months to transform. The majority are stuck in pilot mode.
Meanwhile, workers with AI skills already command 23% higher salaries. That gap is widening every quarter.
You have time to adapt, but start to upskill now if you’re able. And make decisions based on data, not the essay going viral this week.
Human+AI is an independent publication by Nicolle Weeks. While I serve as Director of AI Communications at Manulife, all views expressed here are solely my own and do not represent the views of my employer.
Did someone send you the “Something Big” essay? Share this post with them.
AI in the news
Elon Musk’s xAI overhauls top management, lays off employees (Globe and Mail) Elon Musk folded xAI into SpaceX, reshuffled its leadership, and watched half of its original co-founders walk away. He’s preparing the $1.25-trillion behemoth for a record-setting IPO and a future of data centres in space. It’s a reminder that the AI race is about power, infrastructure, and who controls the systems shaping our economy.
TikTok creator ByteDance vows to curb AI video tool after Disney threat (Guardian) ByteDance’s Seedance 2.0 demo wowed users by spinning simple prompts into cinema‑style clips, but it quickly ran into trouble when studios accused it of training on their characters and demanded safeguards to stop copyright theft. The backlash exposes the tension between our fascination with generative video tools and the growing call for clear, fair rules that honour artists.
AI is giving you a personalized internet, but you have no say in it (NYT) Google and Meta are rebuilding the internet around AI agents that summarize your emails, reshape your search results, and tailor ads and prices to what you reveal in conversation, often without a clear off switch. The result is a hyper-personalized web that feels helpful on the surface, but shifts power toward platforms and advertisers, leaving most of us with less control over how our data, attention, and wallets are being shaped.
Thank you for this cohesive, clear breakdown! I read Schumer's article as well but it left me wondering what 'viral' even means today. Did he just get a bunch of bot farms to boost this article? That's where we are - but your breakdown of what's real and what isn't is grounded.
Thank you.
Totally agreed.
Work will evolve a lot, and yes; it will happen *relatively* quickly - but it's going to be a lot messier and a lot more uneven than many would want us to believe.
In general, it's very important to be sceptical of anyone who claims that such-and-such outcome is "100% inevitable" because *inevitably* they're thinking about the problem from within a highly bounded frame of reference. Within that frame, their logic may be perfectly sound, and that's why a nicely written essay can very coherently and completely support their position.