---
id: "lib-2026-09-00-ai-adoption-in-chinese-enterprises"
type: "library"
title: "AI Adoption in Chinese Enterprises"
lang: "EN"
org: "ILO"
source_type: "Article"
date: "2026-09-01"
week: "2026-W36"
url: "https://nosov.ai/library/lib-2026-09-00-ai-adoption-in-chinese-enterprises/"
original_url: "https://www.ilo.org/sites/default/files/2026-09/Research%20Brief_Artificial%20Intelligence%20Adoption%20in%20Chinese%20Enterprises.pdf"
tier: "A"
tier_name: "Strong evidence"
tier_reason: "ILO and Renmin University researchers combine 21 structured firm interviews with a survey of 1,591 Chinese professionals, rare first-hand evidence on enterprise AI in China, though productivity figures are self-reported."
summary_basis: "full-text"
original_minutes: "25"
tags: ["productivity", "hybrid work"]
frames: ["solow-paradox"]
---

# AI Adoption in Chinese Enterprises

Tier A (Strong evidence): ILO and Renmin University researchers combine 21 structured firm interviews with a survey of 1,591 Chinese professionals, rare first-hand evidence on enterprise AI in China, though productivity figures are self-reported.

Summary based on the full text.

## In short

An ILO research brief (September 2026) by Ekkehard Ernst (ILO) with Zhong Zhao and colleagues at Renmin University of China. Interviews with 21 Chinese firms, from an 8-person media startup to a 270,000-employee insurance group, find that every firm uses AI or plans to soon. Where gains are measured they are large, but most firms lack any systematic way to measure them. Firms run AI as human-AI collaboration rather than full automation, and workers are split between optimism and income anxiety.

## Key numbers

- 6,000 → 15,000: daily customer issues handled by 300 insurance service staff after AI deployment (+150%)
- 30 → 13 days: recruitment cycle time at a large insurance group (-57%)
- 39%: surveyed professionals expecting AI to lower their income
- 47%: surveyed professionals who believe AI creates more jobs than it displaces

## What it reports

- All 21 firms reported active AI deployment or concrete plans; sectors include manufacturing (6 firms), finance, insurance and real estate (4) and business services (4).
- A survey of 1,591 professionals: 56% see AI as an inevitable trend, 47% think it creates more jobs than it displaces, about one-third worry about social stability or human uniqueness, 39% expect income declines.
- Self-reported gains: 30% production efficiency at a "lighthouse factory" (about 20% at other client factories); a specification-reading team cut from over 10 to 2-3 people; changeover time from 120 to 60 minutes; labour costs down about 80% in specific scenarios.
- Services: 300 insurance agents went from 6,000 to 15,000 daily issues; recruitment cycle from 30 to 13 days; project managers from 2-3 to 5-6 simultaneous projects; an HR services firm cut R&D hours 30% and raised per-capita service output 28%.
- A talent firm says AI lets junior consultants deliver work that needed mid-to-senior staff, which the authors contrast with US concerns about disappearing entry-level jobs.
- Three adoption models: centralised specialist teams, business-embedded integration and bottom-up diffusion; most firms are at the tool or process stage, few at business-model change.
- Barriers: output quality (8+ firms, including customer-service accuracy below the 95% needed for direct client use), skills and resistance (6+), regulation and data security (6+), integration (5+), cost and ROI (5+). Staff over 40 show weaker AI skills.

## How they got it

Structured qualitative interviews with 21 enterprises across sectors, ownership types and sizes, run by Renmin University researchers, covering motivations, applications, measurement, challenges and investment plans. A separate quantitative survey of 1,591 professionals across industries, ownership types and career levels captured views on jobs, income and work organisation.

## Limits

- All productivity figures are self-reported by firm representatives; most firms have no formal measurement.
- 21 firms is a purposive sample chosen for variety, not representativeness.
- Survey method details (sampling, weighting) are not given in the brief.

## Why it matters for companies

The Chinese firms describe the same pattern seen in the US and Europe: big local gains in customer service, recruitment and document work, and little firm-wide measurement. The brief shows practical controls, such as systems that withhold output unless source documents were consulted and human review below a 95% accuracy threshold. HR functions appear both as heavy AI users and as the staff most likely to resist it. For multinationals with Chinese operations, it shows local tools such as Doubao, Kimi, Yuanbao and Feishu in daily use alongside Copilot.

Original: https://www.ilo.org/sites/default/files/2026-09/Research%20Brief_Artificial%20Intelligence%20Adoption%20in%20Chinese%20Enterprises.pdf