Early adopters moved on. They know what's out there and shop around for the best tools
ChatGPT's moat is weaker than it looks. Power users churn out while mainstream users churn in
Mass adoption is about utility. Less exploration, more immediate results
2/ ChatGPT writing use peaked & plateaued
Writing is still the top use case (40% of work messages), but growth flatlined in 2024. Technical help is trending down, while info-seeking and multimedia are on the rise.
AI writing hit a ceiling (for now). People see the sameness and value their own voice more
Technical users are fragmenting to specialized tools. Early adopters who used ChatGPT moved to other tools over staying with the generalist one
3/ Automation is overtook augmentation
Anthropic data shows automation (AI completing tasks with minimal input) passed augmentation (human-AI collaboration). Users are iterating less and delegating more.
Knowledge work is already shifting. The most educated workers are deep into AI, the others will follow
Institutions lag behind individuals. Top users push adoption faster than their orgs can adapt
The productivity gap may mirror the education gap. Workers who master AI will increasingly outperform those who don’t.
5/ AI boosts ideas, but drains wellbeing
AI-assisted groups generate 44% more novel ideas. Yet 82% of users report lower job satisfaction from reduced creativity and skill use. Exposure to AI suggestions also increases conformity and discourages debate.
