Journalism

Q: What have you learnt about using AI in journalism?

A: While artificial intelligence has been making headlines as a potential threat to journalism, the reality in newsrooms today tells a different story. After years of intensive experimentation with AI tools, clear patterns have emerged about what works, what doesn’t, and how publishers can responsibly harness AI to strengthen rather than supplant journalism.

The first and perhaps most crucial learning has been about speed and workflow integration. AI tools must operate faster than traditional manual processes, or journalists simply won’t use them. Early experiments with standalone AI applications largely failed because they added extra steps to already complex workflows. Success came when AI was embedded directly into content management systems, allowing journalists to enhance their stories without switching contexts or interrupting their creative flow.

A second key insight has been the importance of matching different AI capabilities to different journalistic tasks. Where AI truly shines is in content enhancement and distribution — automatically generating metadata, optimising SEO, and adapting stories for different platforms. These tasks, which once consumed hours of valuable journalistic time, now happen almost instantaneously. However, when it comes to actual content creation, the learning has been clear: AI works best as an enhancer rather than a creator, helping journalists brainstorm angles or suggest additional research areas while leaving the core storytelling firmly in human hands.

The governance lessons have been equally valuable. Publishers who successfully deployed AI shared one common trait: they put editorial leadership firmly in control of the technology. This meant creating clear policies about where AI can and cannot be used, establishing editorial control over prompt libraries, and developing training programs to ensure journalists understand both the capabilities and limitations of their AI tools. This necessitates tools that provide such governance functionality.

Perhaps the most surprising learning has been about the human factor. Far from resisting AI, journalists have embraced it — but selectively. They’ve shown enthusiasm for delegating technical and repetitive tasks to AI, especially those that have proliferated in the digital age. Tasks like creating multiple headline versions for A/B testing, generating social media variants, or optimising content for different platforms are now routinely handled by AI. This has allowed journalists to focus more intensively on the aspects of their craft that require human judgment, creativity, and ethical decision-making.

Integration success stories, particularly from European publishers we work with, have revealed another crucial learning: the importance of measurable outcomes. When AI tools demonstrably improve article discovery, increase reader engagement, or reduce time spent on technical tasks, adoption follows naturally. These publishers have seen dramatic improvements in workflow efficiency, with tasks that once took hours now completed in seconds, while maintaining or even improving content quality.

As we guide publishers through this transformation, these learnings point to a clear conclusion: AI in journalism isn’t about replacing human judgment or creativity. It’s about amplifying journalistic capabilities by handling the mechanical aspects of digital publishing that have increasingly burdened newsrooms. When implemented thoughtfully, with clear governance and a focus on genuine workflow improvements, AI allows journalists to do more of what they do best: telling the stories that inform, engage, and move their audiences.

Q: In which use-case has AI proved most effective?

A: One of our forward-thinking clients identified the potential of AI early on. Even before the emergence of generative models like ChatGPT, they utilised traditional AI approaches such as entity recognition and content recommendation to empower their editors and streamline content creation.

When generative AI arrived, they swiftly integrated ChatGPT into their workflows, collaborating with us to achieve the most effective AI integration possible. This initiative was grounded in a clear understanding of efficient workflows and editorial management. The initial focus was on transforming press releases into fully-fledged articles, a task that previously demanded considerable editorial time. By automating this process, they not only accelerated content production but also maintained high-quality standards.

Subsequently, they expanded the use-case to include the automated generation of metadata, social media posts, and search-optimised messages. This comprehensive approach ensured that content was not only created more efficiently but also distributed effectively across multiple platforms. The results were significant: our client was able to increase both their editorial output and the overall quality of their content.

Three best practice top tips

1. Collaborate for optimal integration. Work closely with AI experts and AI-optimised tools to integrate AI solutions that align with your specific needs.
2. Start with clear use-cases. Begin by implementing AI in well-defined areas that offer immediate benefits, such as automating the creation of articles from press releases.
3. Maintain editorial oversight. Establish clear policies and guidelines to keep editorial leadership in control of AI usage.

Purple, a Berlin-based digital publishing platform, supports over 600 brands worldwide. Its core modules, Purple Editorial and Purple Experience, offer AI-driven CMS, multichannel content distribution and enable seamless, consistent user experiences. With products like Purple Essentials and Pro, Purple empowers efficient, multibrand management, making it ideal for publishers of all sizes.

Email: contact@purplepublish.com

Website: www.purplepublish.com