Heavy weekly use more than doubled in six months, a measurable signal of demand growth that affects inferenceRunning a trained model to get answers — the phase where AI is actually used, as opposed to trained.Full definition → load and the market for AI-native products.
Key takeaways · AI-distilled
Any weekly AI use rose from 50% of US adults in March to 56.4% in August, the 2-5 day group grew from 24.3% to 27.3%, and the share reporting no AI use fell from 50% to 43.6%.
Both waves ran on Ipsos's probability-based KnowledgePanel and were weighted to represent US adults; all shares are of US adults, not of AI users.
Epoch flags a method change: March asked one overall days-of-use question, while August asked day counts for each service and collapsed them into the same 1, 2-5 and 6-7 day groups.
The 90% intervals for near-daily use do not overlap: 7.2-9.3% in March versus 16.7-20.8% in August.
Terms in this piece · Glossary
inference — Running a trained model to get answers — the phase where AI is actually used, as opposed to trained.
