Japan faces labour shortages, an ageing population and long-standing productivity problems, yet businesses in the country remain relatively slow to adopt artificial intelligence (AI) compared with those in the United States and the United Kingdom.

Only 8.4% of Japanese workers use AI at work, according to an OECD report published late last year. The figure compares with 50% in the US and 32% in the UK. In Singapore, 56% of workers reportedly use AI several times a week.

So why is Japan moving so cautiously?

Experts point to a combination of corporate risk aversion, concerns over errors, limited digital skills, outdated computer systems and a reluctance to replace human workers with AI.

Austin Xu, co-founder of US start-up Kuse AI, which recently opened an office in Japan, said the conservative nature of many Japanese companies is one reason for the slow adoption.

“Some organisations genuinely take a long time to make decisions because of their process and consensus-driven culture,” he said. “Tolerance for AI mistakes is close to zero, particularly in client-facing roles. Some would rather leave a position vacant than allow a machine to handle it.”

Xu said the approach differs from that in the US, where some companies are giving AI agents greater freedom to improve productivity. American firms often introduce AI as an assistant and gradually integrate it into workflows while managers test the technology and address mistakes.

Japanese companies, by contrast, often demand more evidence before trusting AI, he said.

Parrisa Haghirian, professor of international management at Kyoto University of Advanced Science, also identified risk aversion as a major factor.

“The challenges of adopting AI are similar to those involved in introducing any kind of change in Japanese firms,” she said. “This is why AI use remains limited and cautious, particularly in workplaces.”

She said Japanese companies are particularly sensitive to errors, uncertainty and potential reputational damage compared with the more entrepreneurial culture in the US.

As generative AI remains imperfect, Japanese companies mainly use it for relatively low-risk tasks such as writing, summarising documents and gathering information. Its use is more limited in core operations, decision-making and broader business-process improvements, Haghirian said.

Japan’s healthcare sector has been particularly slow to embrace AI. Some hospitals have yet to fully digitise patient records.

“Paper documents accumulate at a staggering scale,” an employee at a Japanese hospital, who requested anonymity, said. “It’s like the Stone Age.”

The Japanese government has recognised the problem and is seeking to encourage greater business adoption of AI.

Last year, parliament passed the country’s first AI Promotion Act, adopting a light-touch regulatory approach aimed at encouraging investment in the technology.

Tokyo has pledged to make Japan “the world’s most-friendly country for developing and utilizing AI”.

The government views AI as a potential tool to boost productivity, an area where Japan continues to lag behind other G7 economies.

Japan’s Ministry of Finance says business adoption of AI has increased substantially, with 75% of companies now using the technology, compared with 11% five years ago.

Critics, however, say the figure may obscure how extensively AI is actually being used within companies. In many cases, only a small proportion of employees use AI, and their use may be limited to a narrow range of tasks.

Prof Yasushi Ogasawara, an expert on Japan’s social system and technology at Meiji University, said Japan also lacks enough workers with strong technology skills.

“Although Japanese people like using gadgets such as smartphones, digital literacy is low here,” he said.

A report earlier this year estimated that Japan could face a shortage of nearly 800,000 IT professionals by 2030.

Many Japanese companies also continue to rely on outdated computer systems, with around 60% reportedly more than two decades old.

Ogasawara said companies also face pressure to avoid contributing to unemployment, making them more cautious about developing AI that could replace human workers.

“The government is talking about AI, reskilling and digital skills, but in reality it is difficult to adapt human resources to digital skills because the top priority is maintaining full employment,” he said.

There are signs of change, however, with some companies beginning to prioritise AI literacy when recruiting new graduates.

Major trading company Kanematsu, for example, is hiring graduates with strong AI skills. Uta Yamaguchi, who joined the company in April, said she frequently uses AI at work, as do many of her colleagues.

She uses AI to draft emails for her boss, summarise complex documents and record and summarise meetings.

However, she said AI systems capable of independently handling multi-step tasks remain virtually unheard of in Japan compared with the US. Her employer has now begun introducing such systems.

Ogasawara said the pace of change in Japan remains limited.

“Japanese society expects painless reform, so I think it can be said that drastic reform is difficult, to say the least,” he said.

While Japan’s cautious approach may limit disruption, experts say a greater willingness to embrace change will be necessary if the country is to secure the productivity gains it urgently needs from AI.