59 Predictions · Live Tracking

Prediction Tracker

Public AI predictions tracked against outcomes. Who called it, who missed, and where the consensus is heading. Everyone predicts. Few keep score. Our methodology.

Correct95%

No artificial general intelligence in 2025, despite claims by Elon Musk

timelinescoreboard
Gary Marcus2025-01-01
Correct95%

Hallucinations continue to haunt generative AI in 2025

capabilityscoreboard
Gary Marcus2025-01-01
Correct95%

Less than 10% of workforce replaced by AI in 2025

laborscoreboard
Gary Marcus2025-01-01
Correct95%

Few if any radiologists replaced by AI in 2025

laborscoreboard
Gary Marcus2025-01-01
Correct90%

The AI culture war begins in 2025

policyscoreboard
Casey Newton2024-12-20
Correct85%

AI agents will be endlessly hyped but far from reliable in 2025, except in narrow use cases

capabilityscoreboard
Gary Marcus2025-01-01
Correct85%

AI model profits remain modest or nonexistent in 2025

economicsscoreboard
Gary Marcus2025-01-01
Correct85%

US will have very little AI regulation protecting consumers in 2025

policyscoreboard
Gary Marcus2025-01-01
Correct85%

Humanoid robotics: lots of hype, nothing close to Rosie the Robot in 2025

capabilityscoreboard
Gary Marcus2025-01-01
Correct80%

OpenAI annualized revenue hits $18B by end of 2025

economicsscoreboard
Daniel Kokotajlo2025-04-03
Correct80%

AI agents will join the workforce in 2025

laborscoreboard
Sam Altman2025-01-06
Correct75%

AI smarter than humans in many domains

capabilityscoreboard
Geoffrey Hinton2023-05-04
Correct70%

Google mostly catches up to OpenAI in LLM quality

capabilityscoreboard
Casey Newton2023-12-20
Correct70%

Romantic relationships with AI companions increase sharply

socialscoreboard
Casey Newton2023-12-20
Correct65%

Meta kills celebrity-based AI characters

socialscoreboard
Casey Newton2023-12-20
Correct25%

X (Twitter) will be merged into xAI

socialscoreboard
Casey Newton2024-12-20
Correct20%

AI becomes a bigger political issue than abortion or climate change by 2028

socialscoreboard
Scott Alexander2023-12-01
Wrong80%

A new crypto meme coin briefly reaches $100B market cap

economicsscoreboard
Kevin Roose2024-12-20
Wrong75%

SWE-Bench Verified reaches 85% by mid-2025

capabilityscoreboard
Daniel Kokotajlo2025-04-03
Wrong60%

Open source will overtake closed source as most-used AI models

capabilityscoreboard
Mark Zuckerberg2025-04-29
Wrong60%

AI-generated sludge outcompetes digital media for ad dollars

economicsscoreboard
Casey Newton2023-12-20
Wrong50%

Apple acquires Snap

economicsscoreboard
Kevin Roose2024-12-20
Wrong20%

OpenAI officially declares they have achieved AGI in 2025

timelinescoreboard
Kevin Roose2024-12-20
Partial70%

The first year of the agentic era mostly disappoints

capabilityscoreboard
Casey Newton2024-12-20
Partial60%

China achieves AI application parity with US by 2025

geopoliticsscoreboard
Kai-Fu Lee2024-09-01
Pending95%

EU AI Act high-risk system compliance enforcement begins August 2026

policyscoreboard
EU Commission2024-03-13
Pending85%

Anthropic will still be excluded from new DoD contracts on July 1, 2026

policyoriginal
Future Shock2026-03-06
Pending85%

16+ becomes the new norm for social media accounts worldwide

policyscoreboard
Casey Newton2025-12-19
Pending85%

AI will write most of Meta's code within 18 months

laborscoreboard
Mark Zuckerberg2025-04-29
Pending80%

Powerful AI could arrive as early as 2026

timelinescoreboard
Dario Amodei2024-10-01
Pending80%

Most claimed AI breakthroughs in social science will not be reproducible

capabilityscoreboard
Arvind Narayanan2024-09-24
Pending80%

AI smarter than the smartest human by 2025 or 2026

timelinescoreboard
Elon Musk2024-04-08
Pending80%

AGI by 2029

timelinescoreboard
Ray Kurzweil2005-01-01
Pending80%

Technological singularity by 2045

timelinescoreboard
Ray Kurzweil2005-01-01
Pending75%

AI creates measurable white-collar productivity gains within 3 years

laborscoreboard
Erik Brynjolfsson2025-01-01
Pending70%

At least one former Anthropic government customer will publicly state they have switched to a competitor by July 1, 2026

policyoriginal
Future Shock2026-03-06
Pending70%

World models will be the dominant AI architecture within 3-5 years

capabilityscoreboard
Yann LeCun2024-11-01
Pending70%

AI will compress a century of biological progress into a decade

capabilityscoreboard
Dario Amodei2024-10-01
Pending70%

Superintelligence possible within a few thousand days

timelinescoreboard
Sam Altman2024-09-23
Pending70%

US will have a Manhattan Project-scale national AI program by 2027-2028

geopoliticsscoreboard
Leopold Aschenbrenner2024-06-01
Pending70%

AI will add less than 1% to US GDP growth over the next decade

economicsscoreboard
Daron Acemoglu2024-04-01
Pending70%

AGI will be achieved within 5 years (by ~2028)

timelinescoreboard
Jensen Huang2023-11-29
Pending70%

Massive knowledge work displacement within a decade

laborscoreboard
Geoffrey Hinton2023-05-01
Pending60%

AI companies train models that begin to automate AI research itself by 2026

capabilityscoreboard
Daniel Kokotajlo2025-04-03
Pending60%

Level 4 autonomous vehicles operating in 10+ US cities by 2028

capabilityscoreboard
Industry Consensus (Waymo, Cruise, Aurora)2025-01-01
Pending60%

Superintelligence could happen this decade

timelinescoreboard
Ilya Sutskever2023-07-01
Pending50%

AGI (country of geniuses) ~50% by 2026-2027

timelinescoreboard
Dario Amodei2026-02-01
Pending50%

By 2027, AI agents can fully automate the work of an AI researcher given appropriate tools

capabilityscoreboard
Daniel Kokotajlo2025-04-03
Pending50%

Full coding automation by 2027

capabilityscoreboard
Daniel Kokotajlo2025-04-03
Pending50%

AGI will emerge in 5 to 10 years (from 2025)

timelinescoreboard
Demis Hassabis2025-03-17
Pending50%

50% chance of transformative AI by 2040

timelinescoreboard
Ajeya Cotra2022-08-11
Pending50%

50% probability of minimal AGI by 2028

timelinescoreboard
Shane Legg2009-01-01
Pending38%

Multi-Domain AGI by October 2027

capabilityoriginal
Future Shock2026-02-27
Pending35%

35% chance of transformative AI by 2036

timelinescoreboard
Ajeya Cotra2022-09-01
Pending30%

International AI governance body with binding authority established by 2028

policyscoreboard
Yoshua Bengio2024-12-01
Pending15%

15% chance of transformative AI by 2030

timelinescoreboard
Ajeya Cotra2022-09-01
Pending10%

AGI will not be achieved by 2029

timelinescoreboard
Gary Marcus2024-06-22
Pending

Taiwan chips are the single biggest point of failure for the world economy

geopoliticsscoreboard
Scott Bessent2026-01-20
Pending

China will be ready to take Taiwan by 2027

geopoliticsscoreboard
Adm. Philip S. Davidson2021-03-09

AI news, analysis, and weekly deep dives. No hype.