Does AI help in customer support mainly cut jobs, or help the people still there get more done?
Does AI assistance typically reduce support staff headcount or increase productivity?
This explores whether bringing AI into customer support and similar work mostly lets companies employ fewer people, or mostly helps the people already there get more done.
This explores whether AI in support work mostly cuts jobs or mostly raises output per person. The direct evidence in the collection points to productivity, not headcount cuts. One caveat first: none of these studies tracks support teams being downsized after AI arrives, so the collection can't settle the headcount side. What it does show is where the productivity gains come from, who gets them, and how much of them is real.
The main study here followed 5,172 support agents at a Fortune 500 company. AI assistance raised issues resolved per hour by about 15% on average Does AI assistance help less experienced workers most?. The surprise is who gained. Newer agents improved in both speed and quality. The most experienced agents got a little faster, but their quality slipped slightly. In effect, the AI passed the best agents' know-how on to the newest hires. That changes the headcount question. The AI isn't so much replacing agents as shrinking the gap between a new hire and a veteran.
To see why jobs don't simply disappear, look at how work is broken down. Narayanan and Kapoor argue that AI speeds up the middle part of knowledge work, the actual doing. Deciding what to do and delivering it to a person stay the same size or even grow Does AI really compress all layers of knowledge work equally?. They point to translation and legal work, where employment has held steady or grown despite big AI gains. A survey of executives points the same way: labor gets moved around rather than cut, and leaders believe the gains are bigger than the numbers show Do AI productivity gains feel larger than they actually measure?.
The productivity gain itself also shrinks once you look closely. Workday found that close to 40% of the time AI saves goes to fixing and checking its output. Only 14% of employees consistently came out ahead, and they tended to work at organizations that retrained staff and redesigned roles Where does AI's time savings actually go in practice?. BCG found that self-reported productivity actually falls once workers juggle four or more AI tools, because overseeing all of them wears people out and raises quit intent Does using more AI tools always boost worker productivity?. So the extra output isn't free. Part of it turns into supervision work, and supervision is something people still do.
The finding you might not expect concerns those newer agents who gained the most. Other studies in the collection suggest that the better performance you get with AI doesn't always build lasting skill. Workers who did well with AI did no better when they later worked alone Does AI assistance help workers learn lasting skills?. Gains appear when people apply skills they already have, not when they're learning new ones When does AI actually boost worker productivity?. That raises a staffing question no study here answers yet. If AI lifts junior agents without turning them into seniors, where will the next generation of experts come from, the people who can spot when the AI is wrong?
Sources 7 notes
A study of 5,172 support agents at a Fortune 500 firm found a 15% average productivity gain from AI assistance, with gains concentrated among less experienced workers who improved both speed and quality. The most experienced agents saw small speed gains but slight quality declines.
Narayanan and Kapoor argue AI narrows only the middle execution layer of knowledge work while decide and deliver layers persist or grow. Translation and legal work show stable or expanding employment despite AI gains, suggesting task-level compression doesn't shrink occupational demand.
A survey of 750 executives found that perceived AI productivity gains exceed measured ones, likely because revenue lags operational improvements. Effects concentrate in high-skill services and finance, with labor reallocating rather than shrinking overall.
A Workday-commissioned survey of 3,200 active AI users found that while 85% save 1–7 hours weekly, almost 40% of those savings disappear into correcting errors and verifying outputs. Only 14% of employees consistently see positive net outcomes, with success tied to organizations that retrain staff and redesign roles rather than simply deploying tools.
BCG's survey found self-reported productivity rose with up to three AI tools but fell sharply with four or more. Workers experiencing this 'brain fry' showed 34% quit intent versus 25% without it, driven by oversight burden rather than tool count alone.
Show all 7 sources
Wu et al. found that workers using generative AI performed substantially better on content tasks, but when performing similar tasks independently afterward, their performance showed no improvement. The capability did not transfer across contexts.
Studies showing AI productivity gains measured tasks within workers' existing domains. When workers used AI to learn new skills, productivity gains disappeared and learning suffered, suggesting prior findings do not generalize to skill acquisition.
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- How AI Impacts Skill Formation
- Toward Measuring AI's Effects on Skill Formation: The Stock-Formation Gap
- What 81,000 people told us about the economics of AI
- Does generative AI narrow education-based productivity gaps? Evidence from a randomized experiment
- Generative AI at Work
- Adoption of Generative AI in the Workplace: Increasing and Shifting the Balance of Productivity and Communication Activity
- Estimating AI productivity gains from Claude conversations
- Beyond Productivity: Measuring the Real Value of AI