Desk Jobs Have Five Years; Trades Get a Decade

If your income depends on a keyboard and screen, your five‑year countdown has started. Tool‑belt trades get just over a decade.

The maths and timelines are brutal, but the changes are real—here’s why the two timers tick at different speeds and how to outrun both for a time.

Don't shoot the messenger; but do shoot me your comments as this will impact us all and especially our children.

The Five‑Year Countdown For Desk Jobs

The Generative Pre-Trained Transformer (GPT) and generative AI is the current wonder technology, but it is probably just a stepping stone to something more powerful and more profound. AI can already write, speak, listen, and create still and moving images. It can communicate in almost every language, and even in some languages that are no longer spoken. It can draft pleadings, write code and now it understands what is on a computer screen and can move a cursor on the screen and type. 'Computer use' is slow now but it will eventually happen at the application or operating system level and not on-screen. In fact, computers won't need screens unless a human wants to observe data or read something. Any job done at a computer will go away.

But, let me remind you. We are experiencing the equivalent of dial-up internet in AI.

Goldman Sachs estimates that 300 million full‑time roles worldwide are “exposed” to automation, largely concentrated in knowledge work sectors such as admin support, law, finance and tech. ( Forbes).

Of course, 'exposure' does not equal instant redundancy, yet corporate roll‑outs follow a predictable S‑curve: early pilots by the leaders, a scramble by fast followers, then a plateau as laggards realise the market won’t wait. However, in an exponential world, playing catch up will be a bad plan. McKinsey has tracked dozens of technologies—from ATMs to RFID tags—and finds the bulk of adoption typically arrives within seven years of a breakthrough’s first commercial splash. 2030 is expected to be the point when desk‑job displacement moves from forecast to fact.

Boards are not hesitating because the tools are weak; they are waiting for liability cover. A large US insurance syndicate recently priced AI‑driven “hallucination losses” at up to US $1.5 million per incident. But, these issues will be engineered out as they have to be to ensure that big tech, venture capitalists and even governments generate a return on their considerable investments. The result: a wave of AI, software and AI agents that are tireless, mistake‑free (enough), and cheaper than the coffee budget—ready to absorb everything that can be achieved with a keyboard, monitor and mouse.

Forget AI. What is the error rate of your coworkers or employees?

What Does the End of the Desk Job Look Like?

Not all knowledge work vanishes at once, it cascades and accelerates over time:

1. Document factories (contracts, document generation, slide decks, most coding, reports, desktop, research) shrink under template‑driven LLM output augmented by vertical AI-driven platforms.

2. Research & analytics follow as agents master data extraction and statistical reasoning.

3. Process-heavy roles—HR compliance, risk QA, payroll, the last coding—lose their differentiator once agents string tasks together end‑to‑end.

4. Creative and strategic functions are last, yet even these are augmented heavily; humans become editors of swarms of AI agents and use AI agents to manage the AI agents, and human resources.

By 2030, companies that cling to purely human workflows collide with three unforgiving forces:

1. Price pressure from automated rivals.

2. Talent drained towards hybrid firms where they are augmented and can focus on more meaningful work.

3. Investors refusing to fund overhead that machines can perform as they become better, cheaper, faster and safer.

How Do Trades Buy Extra Time?

If AI agents scale like software, why can’t humanoid robots do the same for physical tasks? Why does physical work get more time?

Robots already out‑lift us; what still slows them is dexterity. Although AI is expected to equal humans across all measures of cognition by 2028, a humanoid that can thread a pipe, crimp a cable, or align a 0.3 mm screw is another story. Fine‑motor end‑effectors, tactile skins, and tool‑switch agility remain expensive and fragile, so mass rollout drags toward the 2040s. Until a robot can clip a cable tie in a cramped loft faster than a human, plumbers and sparkies keep their decade‑long head start.

Atoms are just slower than bits. Robots have to be built, deployed, maintained and upgraded. NVIDIA's founder and CEO, Jensen Huang, is already selling the era of 'physical AI'.

• Robot supply remains a niche. The International Federation of Robotics counts 4 million industrial robots in service and 541,000 new units shipped in 2023—a record, yet still a rounding error compared to what is coming.

• Actuators need magnets. High‑torque servo motors rely on neodymium and dysprosium; Beijing’s new export permits are being rationed company‑by‑company, underscoring that supply is geopolitical, not just logistical. The speed of the disappearance of human trades may be directly linked to the global trade in the materials needed to build robots, motors and batteries.

• Batteries advance by doubling, not tripling. Even optimistic chemistry road‑maps suggest density gains of 8‑10 % a year—fine for EVs, marginal for a 90‑kg robot that must lug its own power pack to a remote job site.

Assume factories 10× their humanoid‑robot output every three years—far faster than any previous industrial scale‑up. The first billion robots still do not clock in before the early 2040s.

Double, rather than 10×, and saturation slips into the 2050s. Meanwhile, the global working‑age population is over 4 billion.

Humans, inconveniently, are still the lightest, cheapest, self‑repairing multi‑purpose machines on the planet. A plumber with a van, a thermos and eight hours of glycogen outperforms a robot that needs battery swaps, millimeter‑perfect vision and a clean, flat surface to stand on.

Thus trades retain pricing power through most of the 2030s—but with a sting in the tail. Everyone will want to learn a trade!

The Labour Glut

When AI empties office floors, office workers, professionals, ex‑developers and ex‑marketers do not evaporate; they are going to retrain. By mid‑2030s construction, repair and logistics firms face a paradox: physical jobs remain, yet applicant lists overflow with overqualified knowledge workers willing to take a pay cut.

Wages may hold steady, but with increased supply and limited demand, average incomes are likely to fall. Wages may not collapse while robot supply remains scarce, but once the hardware finally scales, and it matches the dexterity of humans, another displacement wave begins washing over physical labour, including the trades.

What Is The Human Moat?

• Licence & liability: Statutes and regulators still demand a human signature on prescription pads, audit letters, legal filings and engineering blueprints. Roughly one‑fifth of the UK workforce—about six million people—work in roles that legally require a professional or statutory licence, giving them a regulatory moat. 24 % of the US workforce holds a licence or professional certification. Until the law changes (and regulators are historically slow), licenced practitioners stay in control.

• Authenticity & empathy: People pay extra for live music, therapy, coaching and storytelling because real human presence feels different from pixels, even photorealistic ones. As synthetic content floods feeds, the appetite for genuine voice and human experiences will grow, not shrink.

• High‑trust negotiation: Enterprise sales, diplomacy and large‑ticket finance hinge on accountability, relationship memory and the subtle dance of concession. A chatbot or robot cannot replicate eye‑contact credibility—at least, not yet.

Cascading Economic Shocks

With cascading automation will come cascading economic and societal shocks. How we all earn a living will be the feature of a future newsletter.

1. Income volatility: The gig economy is coming for all of us as we are employed to deliver discrete tasks or projects and form transient teams. There will be a rise in gig‑style task markets that will swing between increasing oversupply (desk work) and medium-terms undersupply (field service).

2. Retraining churn: The need to retrain and perhaps earn much less will become acute. Governments that subsidise only front‑end “skills boot camps” discover workers must re‑upgrade every three to five years, which will be unaffordable for millions of former keyboard-based workers.

3. Local inequality widens: Regions rich in data centres and robot manufacturing boom; those reliant on clerical back‑office hubs may face decline.

Threaded through all this is the safety‑net debate: negative income tax, universal basic dividends, employee share schemes supported by automated productivity gains.

Crucially, the timeline buys legislators time—they see the wave building across a decade, or perhaps 15-years. No longer.

What Is The Best Personal Strategy For The Next Decade?

1. Own a moat, not just a skill: Letters after your name buy statutory leverage. If licensing is feasible in your field, start the coursework now.

2. Acquire equity in automation: Whether via employee‑share plans, venture or angel investment tickets or fractional ownership of robotic fleets, flip disruption from threat to dividend.

3. Bridge silicon and steel: Pair AI‑ops fluency with a hands‑on trade or domain credential. The human who can talk to both the software agent and the site foreman becomes the indispensable integrator.

4. Compound relationships: AI can scale information but not trust. Networks of mentors, clients and collaborators buffer earnings when job categories wobble.

5. Invest in meaning: Creativity, narrative and values do not get cheaper when code gets better; they become rarer and therefore dearer. Build a portfolio—podcast, community, craft—that showcases an irreplaceable point of view.

The Takeaway

There is no point dancing around these futures. Individuals, organisations, governments should start preparation now.

History rarely offers such clear signposts: five years to re‑engineer keyboard‑based careers, about ten before the spanner comes under equal pressure, and a generation before humanoid robots outnumber us on the shop floor.

So, where does that leave us?

Humans still own trust, empathy and narrative. Get licensed where you can, hold equity where you can’t, and master both the language of silicon and the feel of steel. Five years, ten years or more—curiosity, not nostalgia, is the safest career plan. Which clock are you on?

The jobs clock is ticking, but anyone who can mix deep expertise with adaptable curiosity will still own the timetable.

Which clock are you on—five, ten, or beyond?

Thanks for reading.

If you have found this article interesting, please share it and let me know your thoughts in the comments.

Follow me on Instagram and X/Twitter.