Journalists in France test Spinoza Project’s first AI prototype launched by RSF and the Alliance
Journalists with 12 French media groups are testing the prototype of an artificial intelligence (AI) tool for the media that respects intellectual property. It has been developed by the Spinoza Project, a joint initiative by Reporters Without Borders (RSF) and the General News Media Alliance (Alliance de la Presse d’Information Générale, the Alliance).
A dozen journalists with media outlets that are members of the Alliance began testing this AI prototype in April with the aim of helping to perfect it. The tool they are testing provides access to detailed, precise and sourced information on climate change issues.
The journalists have varied profiles. Some are general news reporters, some are specialists, some work for local media and some work for the national press. The variety is designed to ensure that this AI tool can adapt to the specific uses and needs of media professionals.
The Spinoza Project aims to develop a generative AI tool with and for the media by exploiting the capabilities of large language models (LLMs) to analyse both media articles and technical, legal, and scientific data linked to climate change. Produced by several “design thinking” workshops, the prototype quickly renders and synthesises complex documents, while providing its sources in order to ensure transparency and, if necessary, allow journalists to verify and expand the information.
By means of regular contact and interviews with the testers, the team supervising the Spinoza Project’s development will keep updating the tool and will compile a roadmap for developing new features.
“The Spinoza Project is a concrete proposal by RSF for putting AI at the service of journalism. In contrast to the big tech corporations that would like to create their products without accountability and beyond the legitimate fears that these technologies occasion, this project shows that journalists and the media can be participants in the artificial intelligence story.”
“This tool shows that publishers and journalists are interested in developing generative artificial intelligence that is reliable, creates value and respects intellectual property. By combining editorial skills and the power of LLMs to analyse vast volumes of data, we increase both journalistic production and information quality.”
Database of 12,000 media articles
The prototype contains five databases whose contents were approved during workshops consisting of journalists and media executives working for the 12 French press groups that are participating. The biggest database contains 12,000 print media stories on climate change from the regional or national press. All of these articles come from the media participating in the experiment and were selected by them alone. By respecting the desire of publishers to make or not make their data available, and by providing them with the means to withdraw if they want, the Spinoza Project radically distinguishes itself from the approach of AI purveyors such as Google or OpenAI and offers a formidable tool for promoting news media diversity and richness.
Other databases consisting of scientific reports by the Intergovernmental Panel on Climate Change (IPCC), French legal texts, reports by France’s Environment and Energy Management Agency (ADEME), and documents from the French government's national low-carbon strategy allow the user to quickly and precisely access information contained in these tens of thousands of pages.