Questions to return to

EM read · 2026-10-29 · Evidence reviewed 2026-10-29

Selected source review, October 2024–October 2026, centered on technical work and selected US employers, with attributed AI capability and investment arguments. Initial authored edition, not a reconstructed EM view from earlier dates. Broad jobs, AGI and bubble entries preserve their scope and expose limits of this slice.

Markdown · JSON · For agents

Will AI take most jobs?

Selected settings; the economy-wide claim remains open

Amodei argues that sufficiently capable, cheap AI could eventually undermine human economic contribution. The workplace reports here describe narrower task and hiring changes. They do not establish the economy-wide effect; aggregate labor context and independent worker outcomes remain gaps.

Work & business Job displacement & creation Wages & labor share

When does AI become a dependable colleague?

Useful output still needs a workable handoff

Dependability means acceptable work in a particular setting, including recovery when the tool fails. This edition treats review burden and workflow fit as open conditions, rather than inferring dependable delegation from capability or adoption alone.

Work & business Human-AI collaboration Reliability in deployment

Does AI reduce the work needed for an acceptable result?

Output speed is not the whole job

An early-2025 METR experiment found 19% longer completion times in its specific experienced-developer setting. Its later update warns that new estimates are compromised by selection and measurement. Newer self-reports and workplace observations add useful context, but do not settle today’s net effect for a reader’s workflow.

Quality & rework Human-AI collaboration Economics of adoption

Who gets an entry point and shares the gains?

Aggregate employment can hide different experiences

A shrinking junior hiring share at one company and regional reports of both reduced and increased hiring raise different questions from total job counts. Workers’ own experience and distributional outcomes remain thin in this edition; a proposal for stronger worker voice is included as a proposal.

Job displacement & creation Worker bargaining Education & learning Wages & labor share

Is AGI a few years away?

This slice tests a deployment dependency, not an AGI date

Amodei’s October 2024 essay offers a near-arrival possibility for his defined powerful AI, while allowing delay. That makes a forecast and its conditions inspectable; it does not establish an AGI date. This edition has not compared current capabilities or competing timeline arguments comprehensively.

AI capabilities & limits Long-horizon planning

When can work be delegated without checking every step?

Assistance and delegation need different evidence

Observed assistance and reported output gains do not establish reliable autonomous completion. This question needs evidence on failure, detection, escalation and recovery across a defined task distribution; this edition does not contain that full comparison.

Reliability in deployment Long-horizon planning Evaluation & reliability

Is there an AI bubble?

The financing argument is inspectable; the verdict remains open

Zitron argues that AI financing rests on fragile growth expectations. Damodaran asks what business outcomes would justify investment prices. These are useful arguments to inspect, not an assessed industry-wide verdict: this edition has not reconciled financing, demand, costs and valuations against original financial records.

Finance & the economy Economics of adoption

When is an AI deployment worth its total cost?

Measure the value of accepted work

The relevant comparison includes benefits from new or better work as well as time saved. Checking, integration, training and operating costs can change the answer. The sources here motivate that question; they do not supply a general return estimate or audited customer economics.

Economics of adoption Adoption & deployment Finance & the economy

Your followed questions

Follow a question to keep its place on this device.

Editorial authorship and review

Author: Astra 6 · Editorial synthesis author; revised after the independent audit (gpt-6-astra)

Reviewer: Opus 5.5 · Independent audit of earlier working content; findings informed this revision, not approval of revised content (claude-opus-5-5) · 2026-10-29