Apocalypse Postponed: Ng Reframes Work Around Breadth

No credible road leads to half the workforce sitting idle while streets erupt. That is Ng central claim, and I find it clarifying. Fear scales faster than evidence, and outsized fear produces outsized mistakes. The sibling conversation carried by Acast strikes the same note: Ng treats the moment as a transition to manage, not an apocalypse to survive. Notably, even early figures shared by camps that amplified the fifty percent warnings fail to show the predicted damage. My street level read: prophets counted shuttered shops, but the actual census of storefronts tells a calmer story.

The mechanism is partial, not total. Automation takes a slice of a role while the remainder stays firmly human. I picture my kitchen. The dishwasher cleans plates, yet I choose the menu, judge freshness, and remember what guests enjoy. The appliance accelerates; responsibility remains mine. Work alongside intelligent systems follows the same split: the system drafts, I direct, and I catch errors. The honest headline is therefore a shifting division of duties inside each job, not disappearance of jobs themselves. The remaining slice still demands people, arguably more attentive ones.

Software offers the clearest preview. Teams once focused on the visible surface now own the whole stack behind it: data flows, safety checks, release routines. Scope widens instead of shrinking. Because systems handle surface building blocks quickly, builders stretch across broader territory. Labor signals collected by CNBC point the same way: hiring for software roles continues rather than collapsing. My reading: stronger tools expand the craftsperson bench. The worker who once polished one screen now tends the entire workshop. Whoever learns that breadth becomes remarkably valuable in the market.

The true shortage sits on the talent side. Ng reports he cannot find enough marketers , salespeople, and operators fluent with intelligent systems. People native to these tools get hired fast. Seats exist; prepared candidates do not. Workforce projections tracked by BLS lean the same direction, with computing employment pointing upward rather than downward. My driving license analogy fits: cars stand ready and roads lie open, yet licensed drivers remain scarce. The answer is training drivers, not banning cars. Upskilling is my keyword here, meaning fresh work habits that embrace the whole role, not merely tool familiarity.

Self compounding superintelligence strikes me as slogan over substance, and Ng agrees. The story claims systems improve themselves at an accelerating clip. Appealing, yet physics disagrees. Without fresh compute servers, new power plants, and long rounds of experiments , progress stalls. My greenhouse analogy: however fine the seed, no harvest arrives without water, light, and space. Figures published by IEA show data centre power appetite jumping through 2025, with bottlenecks forcing a scramble for answers. Intelligence growth is therefore concrete and patience bound, tied to buildings, cables, and waiting periods. Until those walls move, compounding talk sounds like raw hype.

The horizon deserves honesty. Ng would welcome systems matching every human intellectual feat, yet he expects several major breakthroughs first. He rules out that leap within months and calls it unlikely by 2030, leaving room for decades of steady advance. I read that as guidance rather than comfort: a long staircase lies ahead, with human effort needed on each step. Until systems cover one hundred percent of human duties, the order of the day is collaboration, not cataclysm. Plenty of work remains, and even more learning.

Entry level anxiety deserves respect. The Stanford study of payroll records, known as Canaries, finds workers aged twenty two to twenty five in highly exposed roles trailing peers by around nineteen percent. The telling detail: the gap comes from reduced hiring , not waves of dismissals. Doors narrow rather than slam shut. Ng blames schools more than systems: some universities still graduate computing students who never once connected to a frontier system, while his teams hire nobody of that profile. Anyone unable to steer an intelligent assistant finds no seat on his crews. His summer interns prove the reverse: school pupils, undergraduates, and fresh graduates fluent with these tools delivered stacks of useful work. Seats exist, and fresh breadth opens them.