RWS survey finds most leaders distrust AI cultural nuance

RWS has published research suggesting that most enterprise content leaders do not trust AI to handle cultural nuance, even as many continue using AI-generated content across global markets.

The findings are based on a survey of 200 senior content leaders in the US, UK and Asia-Pacific who oversee content strategy, localisation and digital communications at large international companies.

The study found that 94% of respondents had little or no confidence in AI's ability to manage cultural and emotional nuance across markets. At the same time, 86% said AI had sped up content creation.

That increase in output appears to be creating strain elsewhere in the workflow. Some 65% of respondents said AI had slowed localisation, while 21% of localisation budgets were being spent on rework to correct or adapt content before it could be used in different regions.

For a multinational business with an annual localisation budget of USD $5 million, RWS said, that level of rework would amount to more than USD $1 million each year.

Cost Pressure

The data points to a widening gap between the use of generative AI and confidence in its results. Businesses may be adopting AI to produce more content more quickly, but the study suggests they are also absorbing extra costs when that material must be rewritten for local audiences.

The issue is most visible in the distinction between translation and localisation. The survey found that 71% of content leaders use generative AI for translation, where the main task is direct language conversion. Only 20% use it for localisation, where content must be adapted to local culture, tone and market expectations.

That difference matters because translation and localisation are often treated as adjacent tasks within global content operations, even though they require different levels of judgement. The findings suggest many organisations are using AI in ways that create more work for human teams once content reaches local review.

Beyond budget pressure, the research also points to operational strain inside content teams. More than half of respondents, 56%, described their organisations as "managing but stretched", while only one in six said they were handling current content demands well.

Confidence Gap

Despite those pressures, the survey found some optimism among respondents. More than half said they expected their organisations to cope better in three years' time, even without major structural changes to the way content is produced and adapted.

RWS argued that such expectations may be hard to sustain if AI-generated volumes continue to grow across more languages, formats and channels. In that setting, errors or awkward phrasing can multiply quickly, especially when content is created centrally and pushed into multiple markets.

The survey highlights a broader question for large companies as they integrate generative AI into marketing, communications and customer content. Much of the debate around AI deployment has focused on speed and scale, but the findings suggest cultural fit is emerging as a separate concern for businesses operating internationally.

For content leaders, that concern is not only whether wording is technically correct. It also extends to whether messaging reflects brand tone, local expectations and emotional context without creating confusion or reputational risk.

Emma Fisher, Vice President of Global Marketing at RWS, said the results exposed a clear contradiction in current enterprise practice. "Ask content leaders whether AI can truly handle cultural nuance, and fewer than one in ten say yes with confidence. Yet the same leaders are scaling AI-generated content across global markets regardless," Fisher said.

She said businesses should focus on improving how AI is used rather than cutting back. "The answer isn't to slow down - it's to deploy smarter AI. AI that understands context, culture and brand intent as fluently as it generates content at scale," Fisher said.