Customer-support productivity uplift from a generative-AI assistant
Change in issues resolved per hour when support agents get a generative-AI assistant, from the Brynjolfsson, Li and Raymond field experiment (NBER WP 31161), with the novice effect and MIT Sloan's restatement shown alongside.
Reason through thisThe reading
A 14% average, 34% for novices and nothing for experts. A tool that levels the floor, not one that replaces the role.
1 reading drawn from nber.customer_support_agents.productivity_uplift.pt, one of 2 series listed under How it is measured; 2 observations in the evidence.
Source: NBER working papers and chapters. NBER.
The numbers
| as of | reading | value | record |
|---|---|---|---|
| 2023-04-30 | customer support agents | 14.0% | estimate |
Why it matters
The best-identified productivity effect on record for a deployed assistant; the normal-technology story predicts modest average gains concentrated in novices, which is exactly what it found.
What cuts against it
One firm, one tool, 2020-21 vintage models; issues per hour is a partial productivity measure; later assistants may do more.
How it is measured
- series
nber.customer_support_agents.productivity_uplift.pt - series
mitsloan.customer_support_agents.productivity_uplift.pt - source MIT Sloan Institute for Work and Employment Research · default tier 6 · MIT Sloan; short quotation
- source NBER working papers and chapters · default tier 6 · NBER; abstract quotation
- proxy deployment, welfare · cadence annual · valve adoption to adaptation
Normal = an average uplift under 20%, the range prior process tools (CRM, knowledge bases) delivered in the same job; fast = 50% or more, the order of magnitude the fast scenario needs from a single tool. Between is `emerging`.
Applied to nber.customer_support_agents.productivity_uplift.pt.
Latest readings
- 2023-04-30 customer_support_agents · 14.0%as of 2023-04-30
Why this status, and its history
Evaluator: 0.14 (NBER working papers and chapters, 2023-04-30) is inside the consistent band (hi=0.2). Auto-reason; band rationale: Normal = an average uplift under 20%, the range prior process tools (CRM, knowledge bases) delivered in the same job; fast = 50% or more, the order of magnitude the fast scenario needs from a single tool. Between is `emerging`.
- 2026-09-10unmeasured to consistent with normalconf — → 75 · the evaluator (automatic)
Evaluator: 0.14 (NBER working papers and chapters, 2023-04-30) is inside the consistent band (hi=0.2). Auto-reason; band rationale: Normal = an average uplift under 20%, the range prior process tools (CRM, knowledge bases) delivered in the same job; fast = 50% or more, the order of magnitude the fast scenario needs from a single tool. Between is `emerging`.
source1
Why this confidence
Rests on 2 observations from MIT Sloan Institute for Work and Employment Research, NBER working papers and chapters; best evidence tier 6, grade B; current.
Confidence is independent of status: 90–95 multiple strong independent sources; 70–89 good evidence, some ambiguity; 50–69 mixed or hard to operationalise; 0–49 limited or vague evidence.