How fast will the impact arrive?
Who says what
Researchers’ documented positions on “How fast will the impact arrive?”
Documented positions only, coded with cited evidence by an AI agent: 16 of 400 mapped researchers so far. Everyone else is unknown: no reviewed evidence, which is neither agreement nor neutrality.
Agree (7)
Christian Catalini agrees
The AGI paper describes the economic transformation as rapid and imminent, stating that the global economy will be transformed over the next one to two years.
Evidence
- “Over the next 12 to 24 months, the global economy will be transformed” — source
Erik Brynjolfsson agrees
Writes that we are approaching transformative AI and that understanding its reshaping of the economy is urgent.
Anton Korinek leans agree
Studies transformative AI timelines seriously enough to model explosive-growth feedback loops and urges preparing now.
Georgios Petropoulos leans agree
He has written that AI-driven disruption may arrive sooner than expected and co-authored a 2021 forecast of a coming productivity boom, while describing a J-curve in which gains follow an initial investment phase.
Gillian Hadfield leans agree
Writes that an economy of AI agents is emerging and institutions must adapt quickly.
Evidence
Michiel Bakker leans agree
Treats AGI-level capabilities as plausible enough on near horizons to require preparedness planning today.
Yoshua Bengio leans agree
The International AI Safety Report update he chairs states that AI capabilities change on a timescale of months or weeks.
Evidence
- “The field of AI is moving too quickly for a single yearly publication to keep pace. Significant changes can occur on a timescale of months, sometimes weeks.” — source
It's complicated (2)
Daniel Susskind it's complicated
Describes technological progress as gradual but relentless and treats tech leaders' five-to-ten-year forecasts of human-level AI as possible rather than inevitable, without endorsing a specific timeline.
Evidence
- “Is that scenario inevitable? No, of course not. But is it possible – I argue that it is.” — source
Diane Coyle it's complicated
Says it is too early to see AI's effects and that institutions change more slowly than the technology, while expecting measurable macro-level productivity gains within two or three years if organizations adapt.
Disagree (7)
Arvind Narayanan leans disagree
Argues AI diffuses like a normal technology — slowly, unevenly, and with institutional friction — rather than transforming the economy within a few years.
Christopher A. Pissarides leans disagree
He has said AI will take a long time to have a real impact and, in 2026, that he does not expect productivity growth to return to past boom levels.
Daron Acemoglu leans disagree
Co-authored vacancy evidence finds AI's aggregate labour-market effects not yet detectable.
Evidence
- “We find no discernible relationship between AI exposure and employment or wage growth at the occupation or industry level, however, implying that AI is currently substituting for humans in a subset of tasks but it is not yet having detectable aggregate labor market consequences.” — source
- “While visible at the establishment level, the aggregate impacts of AI-labor substitution on employment and wage growth in more exposed occupations and industries is currently too small to be detectable.” — source
David Autor leans disagree
Co-authored vacancy evidence finds AI's aggregate labour-market effects not yet detectable.
Evidence
- “We find no discernible relationship between AI exposure and employment or wage growth at the occupation or industry level, however, implying that AI is currently substituting for humans in a subset of tasks but it is not yet having detectable aggregate labor market consequences.” — source
- “While visible at the establishment level, the aggregate impacts of AI-labor substitution on employment and wage growth in more exposed occupations and industries is currently too small to be detectable.” — source
Jakub Growiec leans disagree
Estimates that transformative AI will most likely be developed within one to four decades, a horizon mostly longer than about a decade.
John Van Reenen leans disagree
He has drawn parallels between AI and electricity and the digital revolution, which did not deliver large economy-wide productivity gains until businesses reorganised work, and suggested the same lag may apply to AI.
Evidence
- “These technologies did not immediately produce large economy-wide productivity gains because businesses first had to change how work was organised.” — source
Neil Thompson leans disagree
His study of computer-vision tasks finds only 23% of vision-task wages are currently cost-effective to automate and concludes displacement will be gradual, citing high costs of reaching required accuracy.