Past predictions

What changed?

A dated claim should not float forever. This ledger preserves what was forecast, who said it, the horizon, the qualification—and the question that deserves a fair update.

A fair test

Reconstruct the claim before judging it.

Definitions drift, and timelines often bundle several milestones. Revisit the original wording, separate plans from deployments and outcomes, then ask what evidence would change the judgment.

012028–33

Sep 14, 2023 · Guest forecast

Support, success, and organizational silos may look different as AI handles more execution.

— Rachel Woods

Original qualificationA possibility about organizational design, not a promise of labor outcomes.

What to check nowDid AI change the shape of the organization—or mainly add another layer of tools?

Return to the original conversation →
022024

Dec 21, 2023 · Guest forecast

A trough of disillusionment would be followed by enterprise adoption, regulation, smaller models, and scientific gains.

— George Kamide

Original qualificationSeveral predictions bundled together; each deserves a separate update.

What to check nowWhich parts happened, which did not, and what definition of adoption are we using?

Return to the original conversation →
032024

Dec 28, 2023 · Host forecast

The tool market would consolidate, startups would fail or be absorbed, and election deepfakes would worsen.

— Jeff Wilser

Original qualificationA dated host forecast, not a documented outcome.

What to check nowSeparate market exits, acquisitions, incident frequency, and real-world severity.

Return to the original conversation →
04Before the 2024 election

Jan 4, 2024 · Guest forecast

Little U.S. reform would arrive before the election, while disinformation actors could remain outside U.S. jurisdiction.

— David Becker

Original qualificationAn expectation about law and enforcement capacity.

What to check nowWhat changed in law, enforcement, platform practice, and actual incidents?

Return to the original conversation →
052029 and 2034

Apr 18, 2024 · Vendor forecast

Humanoid robots could reach narrow mainstream use, then broader human capabilities.

— Kordel France

Original qualificationA staged vendor forecast; demonstrations, shipments, and useful work are different milestones.

What to check nowAudit narrow deployment before asking whether broad capability arrived.

Return to the original conversation →
062029–34

Jun 20, 2024 · Vendor forecast

Grounded systems could reach 98–99% accuracy while high-stakes use would still require checks.

— Neil Katz

Original qualificationThe accuracy metric was not fully defined; the safety caveat is part of the claim.

What to check nowDefine the benchmark, error classes, calibration, and consequence of the remaining failures.

Return to the original conversation →
07Before 2027

Oct 14, 2024 · Vendor forecast

A robotaxi priced below $30,000 would become available.

— Tesla

Original qualificationPrice, consumer availability, approved service, and safe operation are separate tests.

What to check nowCompare the plan with delivered vehicles, operating service, approvals, and safety evidence.

Return to the original conversation →
082030

May 8, 2025 · Guest forecast

Many people would stop using accountants and lawyers for routine work.

— Matt Britton

Original qualificationA futurist forecast; task delegation and professional displacement are not the same measure.

What to check nowWhat work was actually delegated—and what happened to usage, errors, regulation, and jobs?

Return to the original conversation →