Our readingWhat is a korrent?
Drawn from what Dario Amodei said
AI and jobs Whether AI displaces human work, which jobs go first, and what replaces them.
On the topicWill AI take the jobs?under our view: Whole classes of work go, and the speed of it is the problem
This is a topic that I warned about very publicly in 2025 , where I predicted that AI could displace half of all entry-level white collar jobs in the next 1–5 years, even as it accelerates economic growth and scientific progress.
Added to korrents 29 Aug 2026 · How quotes work · Something wrong? Tell us
Do you hold this korrent?Do you also believe this?
I hold this tooCannot both be true
Entry-level white-collar work is where junior programmers are hired. If half of it goes within five years, the industry is not simultaneously still hiring the juniors who become the seniors.
Cannot both be true
Most current predictions about AI's imminent impact on the job market are naive.
Karpathy’s claim is second-order: it is about forecasts like this one, and it calls them naive. Amodei’s one-to-five-year displacement forecast is squarely the genre under attack, so the two men cannot both be reading the evidence correctly.
In tension with
AI does not appear to be driving the recent decline in the U.S. college wage premium.
Cowen, looking at the falling college wage premium, says AI is not the main cause. Amodei says half of entry-level white-collar work could go within one to five years. Both can hold — the displacement Amodei forecasts has largely not happened yet — but Cowen is reading the first data that would show it beginning and finding nothing.
In tension with
Predictions that AI would eliminate radiology jobs were wrong; radiology is growing.
Radiology is the standing counter-example to confident displacement forecasts, and Amodei’s is the current confident displacement forecast. The disagreement is about whether this time the forecast is better founded.
In tension with
Amodei reads the labour market as being reshaped by AI. Hansson reads the same layoffs as pandemic over-hiring correcting itself with AI supplied as the cover story. Both can be true at once — a forecast about the coming years is not refuted by a different cause for this year’s cuts — and the disagreement is about which one the current evidence supports.
In tension with
Huang expects NVIDIA to employ more software engineers rather than fewer, on the grounds that a job and the tasks it uses are related but not the same. Amodei expects half of entry-level white-collar work to go within five years. One company can grow while a category shrinks, so this is not a contradiction — but Huang is offering his own headcount as evidence about the category.
Supports
The case for long-term income support rests on the displacement being real and large enough to permanently depress demand for labour.
Shared subject first, then similar wording. They may agree or disagree.