What I’d Teach If We Took AI, and The Future of Work, Seriously
A few years ago, a prestigious university in London asked me to design a course and teach it for a year.
The kind of place that sounds impressive on LinkedIn.
I said no.
I was too deep in the trenches building my business.
But I’ve been thinking lately, if I had said yes, what would I teach now?
Because here’s the thing…
Universities are preparing students for a world that doesn’t exist anymore
By 2030, AI is expected to add $15.7 trillion to the global economy.
That’s more than the current GDP of China.
Yet, universities are still teaching:
How to write a 10,000-word essay
How to pass memory tests instead of executing real work
How to “network” at career fairs
Meanwhile, the real world rewards:
Storytelling
Shipping fast
Owning audience and distribution
AI builders, not users
Most students are graduating with degrees, but no leverage.
They’re job-ready on paper, but irrelevant in practice.
The very idea that we’ll move from learning > working > retiring is obsolete.
And universities haven’t caught up. We hear a lot about how AI will drive economic growth. We also hear about worker empowerment, job creation, and strengthening the middle class. But the reality is, AI could do the exact opposite.
It could hollow out the middle class and shatter the path for those hoping to join it. Because the truth is, AI isn’t just replacing repetitive tasks.
It’s beginning to replace judgment, creativity, and execution, at scale.
Entry-level jobs are the first to go
According to the CEO of Anthropic, 50% of entry-level white-collar jobs are on track to disappear.
We’re already seeing it: entry-level analyst roles are down 30% year over year.
Because AI doesn’t sleep. It doesn’t ask for raises. And it costs less than your monthly coffee budget.
So ask yourself:
Why would a company hire you when a model can do 80% of your work in seconds?
The World Economic Forum projects that 83 million jobs will vanish by 2027, most of them repetitive, entry-level roles. Yes, 69 million new ones will be created. But they’ll require skills most people aren’t being taught.
Jobs that used to be a starting point, think assistants, support staff, social media interns, they’re all being automated.
A McKinsey report confirmed it: the most automatable tasks are the ones junior employees used to handle. In finance, AI builds reports faster than a fresh grad. In media, one person using AI can now do the work of an entire entry-level content team.
Startups are scaling with smaller teams. Corporates are cutting costs by flattening the bottom of the org chart.
In other words, the “entry-level job” is becoming less of a role and more of a signal:
Can you contribute from day one? Can you operate in a world of AI-accelerated work?
The implication is clear. In a market increasingly shaped by automation, the path in is no longer about what you studied, it’s about what you’ve built.
AI companies are telling us what they’re building. Are we listening?
If you’re building a frontier AI model, like OpenAI, Anthropic, or Google DeepMind, it’s almost expected that your messaging leans into big ideas like artificial general intelligence (AGI) and “human-level” capability. That’s how you attract attention and capital.
But hype aside, many are explicitly stating what they’re building: systems that can complete any task a knowledge worker can do.
We should take that seriously.
Economist Anton Korinek, for example, has modelled scenarios in which AGI doesn’t just displace workers gradually, it collapses wages, concentrating value in the hands of AI system owners while leaving non-owners economically sidelined.
He outlines a chilling but plausible arc: wages rise in the short term, but fall sharply before full automation is even achieved.
This isn’t just about job loss. It’s about who controls economic power in a world run by algorithms.
100 days to $100k: the course I’d teach now
This wouldn’t be a class. It would be a sprint you take during your semester.
One thing I’ve noticed is that even the most educated people, the ones with fancy degrees, trophy job titles, or millions raised, often have no idea how to generate $100K in the real world.
Which is wild, because $100K isn’t a lot.
In this course, you won’t just learn. You’ll build something from scratch.
And you won’t do it alone. You’ll have AI as your co-pilot, and insights from top creators, engineers, and operators who’ve done it before.
Here’s how we’d break it down, week by week:
WEEK 1-2: Choose your game
Goal: get clear on your edge. Validate a real problem. Launch in 48 hours.
What You’ll Do:
Identify your unfair advantage (skills, experience, curiosity)
Use AI to generate, cluster, and score business ideas
Validate your idea with real users
Build an audience engine from Day One
Tools:
ChatGPT / Claude → Generate 20 niche ideas, analyse competitors, create positioning maps
Airtable + GPT Plugin → Score ideas based on demand, ease, and differentiation
Tally / Typeform → Build validation surveys
Beehiiv / Substack → Start audience capture through a build-in-public newsletter
Keyword Insights / Glasp / Perplexity → Find what people are searching for and talking about
WEEK 3-4: Build in public like a media company
Goal: create momentum. Turn whatever you build into content. Build trust and leverage.
What You’ll Do:
Set up a Build Tracker to publicly share your journey
Repurpose every task into content (LinkedIn, TikTok, Youtube, etc…)
Create a media engine that builds trust and demand
Tools:
Notion / Airtable → Build Tracker (tasks, assets, launches, learnings)
Descript / Opus / CapCut → Edit short-form content fast
ChatGPT + PromptLayer → Write hooks, headlines, scripts, and captions
Later/Social Bee → Schedule content across social media
AutoPod (for long-form > short-form automation)
WEEK 5-6: Build and launch your first offer
Goal: ship something small and valuable without overthinking.
What You’ll Do:
Turn your idea into a clear offer
Price for value, not effort
Sell before you build (MVP mindset)
Tools:
Canva + Gumroad / Komi → Build landing pages & sell digital products
Notion + Super.so / Typedream / Framer → Launch micro-sites in hours
Stripe + Notion automations → Client onboarding, CRM, receipts
Make / Zapier → Connect payments to delivery (email, digital access)
ChatGPT Custom GPT → Offer validation scripts, pricing calculators, client onboarding flows
WEEK 7-8: Sell before you scale
Goal: manually close customers and build systems that sell.
What You’ll Do:
Build a qualified lead list
Write cold outreach that converts
Use video and automation to scale your pitch
Tools:
Apollo / Clay / Instantly.ai → Scrape leads, qualify, outreach
ChatGPT + Claude → Write custom emails based on job title + pain points
Loom + Tella → Send async sales walkthroughs
CRM in Notion or Folk.app → Track and refine sales pipeline
Warmly.ai → Auto-personalized cold DMs
Make / Zapier → Auto-sequence follow-ups
WEEK 9-10: Automate, delegate and multiply
Goal: build systems and a ghost team to scale
What You’ll Do:
Automate what drains you
Create your first SOPs (Standard Operating Prompts)
Train your first AI teammate
Know when to bring in humans
Tools:
Make / Zapier → Automate workflows across content, sales, onboarding
ElevenLabs / HeyGen / RunwayML → AI for video, voice, and repurposing
PromptLoop / Superflows / Loom AI Copilot → Custom GPT workflows
Fiverr + Upwork → Hire human support for specialised work
Custom GPTs → Create role-based agents: content manager, customer support, researcher
WEEK 11-12: Leverage what you’ve built
Goal: turn your work into momentum. Layer brand, IP and growth.
What You’ll Do:
Productise your process (template, course, micro SaaS)
Package your story into investor or brand-ready decks
Build distribution partnerships
Use audience + traction to unlock new capital (advisors, equity, or investment)
Tools:
Stripe + Zoom → Teach your frameworks, turn them into scalable assets
Skool / Circle / Gumroad → Sell community or bundled IP
Canva → Create brand decks, investor materials
Replit + Claude → Build micro SaaS or apps without coding
But still, there’s a problem with a career built on vibes
There’s a dangerous loop forming: the more you outsource thinking to AI, the less you practice it yourself. Skills atrophy. Judgment weakens. The tool becomes a crutch.
In education, this is even more urgent.
If AI can write the essay, what do we now test for?
Maybe deep thinking, synthesis and critical analysis.
School needs to get harder. More real. More applied.
So here’s what needs to change…
To bridge this gap, universities must radically rethink how they prepare students.
Industry-Integrated Labs: not lecture halls but real-world partnerships where students solve live problems and ship real outcomes.
Access to Builders, Not Just Professors: bring in AI engineers, product designers, operators. People who use these tools daily, not just theorise about them.
Build-First, Not Theory-First: make execution the default. Let students fail, iterate, and try again, with tools like GPT, Midjourney, Figma, and Replit.
New Metrics of Success: move beyond grades. Reward proof-of-work (MVPs, revenue, traction, iteration speed).
AI has lowered the cost of execution but it’s raised the bar on what humans need to bring to the table.
We’re no longer competing on who can do the work. We’re competing on who can think clearly, act decisively, and adapt rapidly in a world where the tools are available to everyone.
The playing field is flatter, but the expectations are higher.
And that’s exactly where most students struggle. Not because they aren’t capable, but because they’ve been trained to optimise for grades, not outcomes. Certainty, not iteration. Obedience, not experimentation.
The students, and future workers, who will thrive won’t be the ones with the highest test scores. They’ll be the ones who think critically, experiment boldly, and learn faster than the market shifts.
That’s the mindset we need.
And that’s the class I would teach.
In Plain Sight is a reader-supported publication. To get the good stuff, consider becoming a free or paid subscriber.