Journalists’ Approaches to Generative Visual AI in Their Work

And the final speaker in this session at the ICA 2024 conference is the wonderful T.J. Thomson, who has explored the use of AI in newsrooms for the past few years in a number of contexts. His present study interviewed journalists at major news outlets in five European countries and Australia, to explore the use of generative visual AI in news production as well as the policies and principles surrounding it.

Key challenges that the (predominantly visually focussed) journalists identified here included the use of AI-generated images to mislead and deceive audiences (for instance in the current war in Gaza); the labour implications of such technologies (a corporate subscription to Midjourney costs about the same as three days of salary for a New York Times photojournalist); the copyright implications of the data that generative visual AI tools were trained on; the lack as well as cost of tools for verifying images as AI-generated or watermarking photos at the point of creation (in cameras) with content credentials; and the brand credibility implications of AI misuse in journalistic coverage.

The use of AI for illustrative, non-photojournalistic purposes is also seen as a potential positive use; so is the use of AI for generating illustration ideas; the cost savings of using AI images instead of expensive stock imagery; the potential for data visualisation through AI; and the use of AI for visualising pre-camera eras.