Sinead Bovell on the Future of Work, AI, and Preparing for Jobs That Don't Exist Yet
A futurist and former model makes the case that we're preparing an entire generation for a job market that won't exist — and the educational system's response to AI is dangerously slow.
Top Claims — Verdict Check
The majority of jobs that today's children will hold haven't been invented yet — and our education system is still preparing them for the old economy
🟡 Partially True“We are teaching kids to compete with machines at the machines' own game — memorization, calculation, repetitive analysis. The jobs that will exist in 2035 require the skills we're not teaching: adaptability, creativity, emotional intelligence, and the ability to work alongside AI. [representative paraphrase]”
AI literacy should be a core curriculum requirement alongside reading and mathematics
🟢 Real“Every child will work alongside AI in their career. Not understanding how AI works — its capabilities, its limitations, its biases — is like entering the workforce in 2024 without knowing how to use the internet. [representative paraphrase]”
The future of work is human-AI collaboration, not human vs AI — but only if we deliberately design for collaboration
🟢 Real“The dystopia isn't AI replacing all jobs. The dystopia is AI replacing jobs that humans need to survive while new AI-adjacent jobs require skills nobody taught them. The gap between old jobs disappearing and new jobs emerging is where the real damage happens. [representative paraphrase]”
The diversity gap in AI development will produce AI systems that don't work for most of the world's population
🟢 Real“The people building AI are overwhelmingly male, overwhelmingly from a handful of countries, overwhelmingly speaking one language. The AI they build reflects those perspectives. If you're not in the room, you're not in the model. [representative paraphrase]”
Young people should be learning to build with AI right now, not waiting for the education system to catch up
🟢 Real“Don't wait for your school to teach you AI. The tools are free, the tutorials are online, and the skills are marketable today. The gap between people who understand AI and people who don't will be the defining inequality of this generation. [representative paraphrase]”
What's Real
The education-workforce gap is documented and widening. The World Economic Forum's Future of Jobs Report 2023 estimated that 44% of workers' core skills will be disrupted in the next five years — the fastest rate of change ever measured. Malaysia's own education data shows the disconnect: universities graduated 50,000+ business administration and management students in 2023, but fewer than 3,000 with AI or data science specializations, despite employer demand surveys showing AI literacy as the number one desired new skill. The diversity gap in AI is backed by hard data: a 2024 Stanford HAI report found that only 20% of AI PhD candidates globally are women, and the geographic concentration of AI talent in the US, UK, and China means that models are trained on data, tested on benchmarks, and evaluated by teams that don't represent 85% of the world's population. The real-world consequences are documented: facial recognition systems show higher error rates for darker skin tones (MIT study), language models perform worse on non-English languages and culturally specific contexts, and recommendation algorithms trained on Western consumer data fail to capture Southeast Asian shopping patterns.
What's Hype
The 'most future jobs haven't been invented yet' claim, while directionally correct, is often used to avoid the harder question of which specific skills to teach. If we can't name the jobs, how do we design curriculum? The vagueness risks producing graduates who are 'adaptable' but expert in nothing — a common criticism of overly generalized education reforms. The counter-evidence from labor economics suggests that most job displacement is role evolution, not role elimination: accountants didn't disappear when Excel arrived; their work changed. The same pattern is likely for AI — most workers will need to learn to use AI within their existing roles, not reinvent themselves entirely. The 'free tools and tutorials' self-education advice, while empowering, overlooks the digital divide: globally, 2.6 billion people still lack internet access, and in Malaysia's rural communities, broadband penetration and device access remain barriers to the self-directed AI education Bovell prescribes. Access to tools doesn't equal access to learning if the prerequisites aren't met.
What They Missed
The economic policy dimension is largely absent. Education reform takes 10-15 years to produce graduates; AI is disrupting jobs now. The transition period requires economic policy interventions — retraining programs, social safety nets, targeted subsidies for industries in transition — that education reform alone cannot address. Malaysia's PENJANA program and HRD Corp's training levy system are existing infrastructure for workforce retraining, but they're not yet calibrated for the AI transition. The talk could be stronger by naming specific policy levers. The employer responsibility is also underexplored: companies that deploy AI to replace workers have an obligation to invest in retraining the displaced — and in Malaysia, the Employment Act 1955's protections around retrenchment create a legal framework that intersects with AI-driven workforce changes in ways that Bovell's Silicon Valley-adjacent framing doesn't address. The ASEAN youth demographic — median age 30, 400 million people under 35, high mobile penetration — represents both the largest at-risk population and the largest potential beneficiary of AI literacy, but this regional context is missing.
The One Thing
The gap between AI displacing old jobs and new AI-adjacent jobs emerging is where the real economic damage happens — and the only way to narrow that gap is to start building AI literacy now, not after the disruption arrives.
So What?
- If you manage a team, allocate 2 hours per week for AI skills development starting this month — the cost of not having AI-literate employees grows every quarter as competitors adopt AI into their workflows
- The diversity gap in AI means Malaysian businesses have an opportunity to build AI applications that serve the local market better than Silicon Valley products — your understanding of local context, language, and culture is a competitive advantage that no foundation model has
- Don't wait for your industry association, government program, or training provider to create an AI curriculum. Start with free resources (Claude, ChatGPT, YouTube tutorials) and build AI skills through actual practice on real business problems this week
Action Items
- 1Start a weekly 'AI skills hour' at your company: every Friday afternoon, one team member demonstrates an AI tool or technique they discovered that week, applied to a real work task. No slides, no theory — just 'here's what I tried, here's what worked, here's what didn't.' The peer learning effect is more powerful than any formal training programme.
- 2Take the free 'AI for Everyone' course by Andrew Ng on Coursera (3 hours total) and require your leadership team to complete it this month. It's designed for non-technical managers and provides the vocabulary and mental models needed to make informed AI decisions. No coding required.
- 3Assess your team's AI literacy today: ask each team member to rate their confidence (1-5) on three things: (1) using a chatbot effectively for work tasks, (2) understanding when AI outputs might be wrong, and (3) identifying one process in their role that AI could improve. Average below 3 on any dimension = immediate training priority.
Tools Mentioned
WAYE (Bovell's organization)
Sinead Bovell's tech education platform — resources on AI literacy and future-of-work preparation for young people
Coursera AI for Everyone
Andrew Ng's free introductory AI course — the best starting point for non-technical teams
HRD Corp (Malaysia)
Malaysia's Human Resources Development Corporation — funds employee training including AI/digital skills, claimable under the training levy
Workflow Idea
Create a 'team AI adoption ladder' with four levels: (1) Awareness — can explain what AI does and doesn't do, (2) User — regularly uses AI tools for at least one work task, (3) Builder — has created a custom prompt, workflow, or automation using AI, (4) Trainer — can teach others to use AI for their specific role. Map each team member to a level today. Set a goal to move everyone up one level within 90 days. Track monthly. This framework turns vague 'we should use more AI' mandates into measurable progress. Most teams discover that moving from Level 1 to Level 2 — just getting people to actually use AI for one real task — produces more productivity gain than any tool purchase.
Context & Connections
Agrees With
- andrew-ng
- kai-fu-lee
Contradicts
- eliezer-yudkowsky
Further Reading
- World Economic Forum Future of Jobs Report 2023 — the definitive data on workforce disruption timelines and skill gaps
- Malaysia's MyDIGITAL Workforce Development Framework — government programmes for digital skills development, including AI literacy