Adoption is real and accelerating — in specific places
By the mid-2020s, several independent lines of research converge on a similar picture: AI use for work tasks has grown fast, and it's concentrated in a fairly specific set of activities — writing and editing, software development, customer service, and data analysis and summarization. Anthropic's own Economic Index, built from aggregated and anonymized patterns in how people actually use Claude for work, has found usage skewing heavily toward these categories, with coding and writing-related tasks among the most common. That's not a guess about the future; it's a real, current signal about what people are already using AI for at work today.
Labor economists studying this, including researchers at MIT and elsewhere, have found measurable effects on specific labor markets: freelance platforms have seen demand shift away from certain routine writing and coding gigs, and some entry-level, routine-task-heavy postings have grown more slowly than comparable roles less exposed to automation. These are real, published findings — not speculation. But it's worth being precise about what they show: shifts in demand and hiring patterns for specific task types, not evidence of mass unemployment or a wholesale collapse of any profession.