The rapidly developing world of AI opens up many opportunities for brands, not least when it comes to content creation and localisation.
But with concerns around accuracy and trust, it is essential that the use of AI tools is based on substantive research and clear frameworks.
Croud recently conducted research into the effectiveness and trustworthiness of AI-generated content, which aims to shed light on the present opportunities and future potential of AI tools. Our research into the use of AI for global content production involved two strands – an extensive consumer survey and an in-depth editorial test.
Our editorial test focused on the accuracy of AI translations. ChatGPT-4 and human translators localised a complex piece of written content into three different languages: French, Spanish and Simplified Chinese. We used a comprehensive scoring system on the translated copy to identify multiple types of error.
The survey results revealed that human translators more accurately translated the piece of content across all three languages, with humans outperforming AI by just 2% in French. However, Chinese showed the largest gap, which can most likely be attributed to the high level of complexity in Chinese translation, with humans outperforming AI by 28% in Chinese.
The second element of our research was a 2,500-person consumer survey. Involving respondents across the UK, US, France, Spain and China, the survey evaluated consumer trust in content produced by AI.
The findings of this research showed a high degree of receptiveness from consumers towards AI-generated content. Seven in ten people shared that they have positive feelings towards AI tools and AI-generated content, with 47% of respondents sharing that they would feel more positively about a brand if they discovered they were using AI to produce content.
Despite these promising findings regarding AI content, a significant portion of respondents across the UK (14%), the US (14%) and France (15%) also shared concerns around the limitations of AI, and described themselves as unlikely, or very unlikely, to purchase from brands using AI-generated content. These concerns are mirrored by the editorial results, which highlighted the existing limitations of AI translation in comparison to human content.
Whilst there are clearly still limitations around AI’s applications to content production and localisation, and human input remains vital, ChatGPT and other AI tools open up exciting possibilities for brands and agencies alike. With the rapid rate of improvement in tools like ChatGPT, it is reasonable to expect the performance in localisation to close the gap on human translators in the near future.
Those companies that embrace these exciting new opportunities that come with AI technology will be the ones that get ahead.