It’s a common misnomer that AI’s efficiencies will reduce the efforts of knowledge workers. In reality, AI doesn’t erase the work; it redirects it.
What used to be hours of drafting is now hours of evaluating, refining, and contextualizing. The real bottleneck isn’t creation anymore; it’s comprehension.
That means reading and synthesis skills won’t decline; they’ll explode in demand. AI can generate a hundred perspectives in minutes, but someone still has to decide which are relevant, which conflict, and which carry risk or opportunity.
Ironically, this backfills demand for all the individuals who felt their “useless” degrees left them behind. Those critical thinking, reading, and interpretation skills aren’t wasted; they’ve become the value-add. In this light, AI doesn’t lower the bar for knowledge work; it raises it.
But here’s the paradox.
AI isn’t just a machine for creating more; it’s a machine for deciding what to leave out. Instead of slogging through 100 documents yourself, you may find AI handing you the three it thinks matter most. At that point, the skill isn’t about reading more; it’s about deciding whether you trust the filter.
Comprehension doesn’t disappear, but it mutates. It becomes meta-comprehension: Do I trust this lens on reality, or do I need to widen it back out?
The Timeline of Reading in the AI Era
- 2010–2022: Pre-AI Baseline: Human effort in reading and comprehension is stable; knowledge workers handle everything themselves.
- 2023–2027: Expansion Phase (where we are today) AI tools amplify outputs dramatically, but haven’t reliably narrowed them. Correction and evaluation demand more human effort, not less.
- 2027–2029: Transition Phase Guardrails and trust layers emerge, reducing noise. Work begins to shift from reading everything to validating what AI prioritizes.
- 2030+: Narrowing Phase AI becomes reliable enough to handle much of the reading itself. Human effort shifts to meta-comprehension: validating and auditing the machine’s filters.
“AI hasn’t reduced our reading load; it’s multiplied it.”
Right now, we’re firmly in the Expansion Phase, and those with critical reading and comprehension skills are in higher demand than ever.
The Third Path: Bifurcation by Domain
There’s also a more complex possibility: not a single trajectory, but multiple curves unfolding by domain.
- High-liability fields (law, medicine, finance): These may remain in the expansion phase indefinitely. The cost of a single misinterpretation is too great, so humans will remain burdened with exhaustive reading, correction, and evaluation long after AI is technically capable of narrowing.
Liability keeps humans in the loop. These domains may never trust AI filters completely.
- Creative and exploratory fields (design, marketing, media): Here, the opposite could happen, workers may leapfrog into the narrowing phase early, trusting AI filters to handle 80–90% of the comprehension load so they can focus on expression, iteration, and innovation.
Creativity thrives on trust. These domains may embrace narrowing faster than anyone expects.
“The future of reading won’t be one curve; it will fracture into many.”
This bifurcation could create fascinating labor market dynamics: one set of professionals drowns in endless reading, while another thrives by outsourcing it almost entirely.
The Cruel Irony
The cruel irony is that the very educational systems that once produced the critical thinkers who excel at reading and synthesis are now under pressure to become more “practical” and “skills-based.”
Yet just as comprehension, interpretation, and reasoning surge back into central importance, we may discover a supply crisis of exactly those skills.
Right when demand peaks, supply shrinks. The pipeline for critical thinkers is drying up just as society needs them most.
“We may run out of critical thinkers just as the world needs them most.”
The Open Question
So the future of reading under AI may not be singular. It may be fractured, expanding, narrowing, or bifurcating depending on the field.
The question isn’t simply: Will AI increase or decrease reading?
It’s: Who will be forced to read more, who will be able to read less, and what happens when the skills required don’t match the supply we’ve built?
“The bottleneck of the AI age isn’t writing; it’s comprehension.”