East and West are heading in opposite directions on AI in research
Western universities are spending fast on AI while their own staff resist it. In Asia the regulator, the employer and the researcher point the same way, and the shift is faster.
The same eighteen months, side by side
| East | West | |
|---|---|---|
| Regulators | Japan’s FSA tells firms not to hold back “out of fear of risk.” Korea’s FSC loosens network separation so firms can connect to AI | AACSB’s 2026 standards for business schools do not contain the word AI |
| Firms | Chinese brokers put DeepSeek-V4 to work within hours. Mirae Asset moves research assistant work to AI | 29% of employees in a US, UK and European survey admit to undermining their company’s AI rollout, and 44% of Gen Z |
| Researchers | Pipelines posted in public on Zhihu, Zenn and GitHub, with token costs and failure lists | Two open letters of refusal, with 2,065 and 1,458 signatures |
| Campuses | No organized faculty refusal turned up in Chinese, Japanese or Korean searches | Faculty petitions against OpenAI contracts at Cal State and Colorado. At Cal State, 16% of faculty finished the training |
| The public | “More likely to think AI has more benefits than drawbacks,” in Ipsos’s words | “More likely to feel nervous about AI.” Among liberal Democrats in the US, concern rose from 45% to 63% in three years |
Ipsos published its fifth annual AI survey on 30 June 2026, covering 32 countries. Its summary puts the regional split in one sentence each: people in Asia and Latin America are on average more likely to see more benefits than drawbacks, and people in Europe and North America are on average more likely to feel nervous.
Western leaders are adapting fast. The people under them are hostile.
| At the top | Among staff and faculty |
|---|---|
| Wharton’s dean took the school all-in in 2024 and promised every student ChatGPT access | One of its research data engineers had told a developer conference not to buy the hype |
| Dartmouth became the first Ivy to adopt AI across the institution, with Anthropic and AWS | Six professors objected on the record, and the faculty union chapter petitioned to remove the provost |
| Harvard gave its members Claude and added a required AI module to first-year writing | Two thirds of 463 faculty say AI had a negative effect on their classes |
| Cal State bought a 17 million dollar ChatGPT license for 22 campuses | 16% of faculty finished the training, and faculty petitioned to cancel the contract |
| Executives in a US, UK and European survey plan layoffs for staff who refuse AI | 29% of employees admit to undermining the rollout, and 44% of Gen Z |
Western universities and firms are spending money and signing contracts. The resistance is underneath, among the people who would have to use the tools.
Asia is shifting faster because the levels agree
In the Asian sources the regulator, the employer and the researcher point the same way. Japan’s regulator tells firms to push ahead. Chinese brokers deploy a new model in hours and publicize it. Researchers post their pipelines for others to copy. No organized refusal turned up in Chinese, Japanese or Korean searches.
Two limits apply. The models themselves are American: Anthropic, OpenAI and Google shipped at least 28 in 2026, and the Asian pipelines here run on Claude Code and Codex. And the Asian evidence is self-reported, with no finance paper and no journal acceptance among it.
The gap is in permission
A Western researcher has the better tools nearby and worse permission to use them. An Asian researcher often has a regulator, an employer and a peer group saying go. Tools cross borders in a day. Permission takes years, and the country that grants it first collects the papers.
- Read what they publish. The Asia section translates it.
- Copy the pattern. A conductor agent, the paper as a set of skills, and a gate on citations and numbers.
- Get your own permission in writing. Make noise while the committee meets.
Sources
- Ipsos, Global Attitudes on AI 2026 ipsos.com
- Financial Services Agency of Japan, AI Discussion Paper version 1.1 fsa.go.jp
- Financial Services Commission of Korea, via the KDI policy archive eiec.kdi.re.kr
- National Business Daily, on DeepSeek-V4 at Chinese brokers nbd.com.cn
- Open letter: Stop the Uncritical Adoption of AI Technologies in Academia openletter.earth
- Pew Research Center, Democrats are now more worried than Republicans about AI pewresearch.org
- CalMatters, on Cal State's OpenAI contract calmatters.org