The AI expectations gap: public and frontier lab views on AI's impact on work
A · Strong evidence Anthropic Institute economists with Charles I. Jones and Anton Korinek compare a weighted Morning Consult sample with an internal staff poll and map answers to a published economic model; strong data, though the staff sample is self-interested.
Summary based on the full text·Original: about 18 minutes to read
In short
A report from the Anthropic Institute (16 October 2026) by Szymon Sacher, Tess Cotter, Charles I. Jones, Anton Korinek and Peter McCrory. It compares a nationally weighted survey of US adults with a poll of Anthropic staff. Both groups agree on what AI does today; staff expect much faster progress, wider use and larger time savings. Both groups link exposure of their own work to higher unemployment in the same way.
Key numbers
- 28% vs 75%average share of own work US workers vs Anthropic staff expect AI to be able to do by 2030
- 39% vs 91%US adults vs staff expecting AI to make a Nobel-level discovery by 2030
- 5.4% vs 17.1%2030 GDP growth implied by the median US adult's vs the median staff member's answers
- n=10,980US adults surveyed by Morning Consult in August 2026, against 1,334 staff
As reported by the source.
English·Web page·opens anthropic.com
What it reports
- 39% of US adults vs 91% of staff expect a Nobel-level discovery by 2030. Between 19% and 40% of adults say AI will never do a given task; at most 2% of staff do.
- Of tasks AI can do, adults expect 42% done with AI by 2030; staff 65%. Adults rank customer preference for people and data privacy as top obstacles.
- 26% of adults who use AI at work say a task takes a quarter of the time or less with AI; 59% of staff. 29% of adult users say it takes as long or longer.
- Share of own work AI can do by 2030: 28% for US workers, 75% for staff; 1 in 20 workers say 80% or more.
- Each 10 more points of own work exposed goes with 0.4 points higher forecast 2030 unemployment, the same gradient for staff. Workers expecting 80%+ exposure forecast 12.2%.
- 95% of staff expect Anthropic headcount to grow; median expected growth 66%. Staff expect a displaced worker to need 12 months to find a new occupation.
- Model-implied 2030 GDP growth: 5.4% for the median adult, 17.1% for the median staff member, 12.5% for 23,075 self-selected website visitors.
How they got it
Morning Consult online opt-in panel, 11-23 August 2026, weighted; an internal staff poll, 9-12 August, unweighted. A September re-run with 7,613 adults changed little. Answers are fed through the authors' "Economic Scenarios for Transformative AI" model.
Limits
- Staff work for a company whose valuation depends on AI optimism; their answers are not a neutral expert panel.
- Online panel excludes people without internet; weighting cannot fix all differences.
- GDP figures depend on the authors' own model and parameter choices.
Why it matters for companies
Executives often plan from either the lab view or the public view. This report shows the gap comes from experience of what AI can do, not from different theories of the labour market, and that regular users move toward the lab view. Staff who use AI daily will expect larger changes to their jobs and to employment; communication and workforce plans should address that, not assume the public baseline.
Summary based on the full text. Read the original for the full argument and data.Read the original on anthropic.com
- Tags
- Future of work, Survey