AI has made the first draft cheap. Drafting, summarizing, structuring and turning notes into usable documents cost little to produce and little to revise, across most common business tasks. Expertise still counts, but it pays in a different place, because the value sits with the AI generalist who owns the outcome from brief to delivery.
In most organizations, the bigger cost is coordination. Splitting a job across narrow roles creates handoffs, rework, meetings and waiting. Each handoff loses context and each review adds delay, which shows up as slower time to market, more management overhead and higher labor cost for the same result.
That squeezes the typical specialist. A worker whose edge is competence in one narrow craft competes with a tool that can match average performance in writing, analysis, design drafts and routine code. People do the best work in each craft. Most specialists, though, have lived in the middle, and that is where AI catches up a little more with every new model.
Three studies put numbers on that shift. A study of an AI assistant in customer support found that it raised the number of issues resolved per hour by 15% on average. The biggest gains went to less experienced staff, so a junior worker started to look more like a solid mid-level one.
Writing and coding show the same pattern. In an experiment published in Science, access to ChatGPT, OpenAI’s AI assistant, cut the time professionals took by 40% and raised the quality of their work by 18%. In a controlled test, developers using GitHub Copilot, an AI coding assistant, finished a task 55.8% faster than those without it.
With the first pass that fast, companies stop paying for it. They pay for setting the intent, shaping the structure and deciding what is true, relevant and safe to ship. That is the AI generalist’s work. Using AI well, one competent person can produce integrated work that would otherwise take several average specialists, with fewer handoffs and fewer delays.
Specialists keep their place where failure is expensive and checking is hard, such as security review, regulated compliance, core architecture and brand-critical creative work. Elsewhere, unless a company can afford world-class specialists in every lane, splitting routine work across roles has become an avoidable cost.
