Adopting Generative AI Tools in Public Service Media Organisations

I’m sharing the next session at the ICA 2024 conference, and it starts with Anis Rahman, whose focus is on the use of AI in public media journalism. AI tools are largely emerging from major corporations in Global North countries; public media organisations are not doing a great deal of work in studying, exploring, or developing AI applications as yet, however. Here it is also important to distinguish between full generative AI tools and mere algorithms.

Uses of AI in newsrooms include direct and indirect uses; tools developed by AI companies and tools developed in-house; uses in pre-production, production, and post-production stages. AI is undeniable and inevitable, and is or will be also in public media newsrooms; but is it in these news organisations best interests to rely on commercially developed tools?

The use of such tools can produce vendor lock-in effects, and especially in a rapidly developing environment this can be highly problematic as better tools emerge onto the market; but public media therefore also find it risky to invest their own development efforts into developing their own AI tools if something better might emerge on the market soon.

There are some exceptions, however: some European public media outlets are doing interesting work. The SWR in Germany requires staff to tag all AI-generated content to maintain transparency and accountability; others are using AI tools in various interesting ways. Training data for AI also comes from news media sources, of course; AI needs journalism for data input, and arguably needs journalism more than journalism needs AI.

The oligopolistic structure of the AI industry, and its lack of transparency and accountability, pose fundamental issues for public media engaging in the use of AI. But conversely there are also problems in trusting and working with local AI start-ups; there is a need for more AI regulation (which does not always serve journalistic interests, however), and greater taxation of AI companies could be used to subsidise public service media.

Non-profit collaboration might also be possible, but this is as yet underfunded and small-scale. Radical approaches, by contrast, would include communally owned and run AI projects; and there may also be an Indigenous approach to AI development – this has been tried in New Zealand with Maori communities, for instance.